South Asia River Flow Projections and Their Implications for Water Resources

icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion Hydrol. Earth Syst. Sci. Discuss., 12, 5789–5840, 2015 www.hydrol-earth-syst-sci-discuss.net/12/5789/2015/ doi:10.5194/hessd-12-5789-2015 HESSD © Author(s) 2015. CC Attribution 3.0 License. 12, 5789–5840, 2015

This discussion paper is/has been under review for the journal Hydrology and Earth System South Asia river flow Sciences (HESS). Please refer to the corresponding final paper in HESS if available. projections and their implications for water South Asia river flow projections and their resources implications for water resources C. Mathison et al.

1 1 1 2 C. Mathison , A. J. Wiltshire , P. Falloon , and A. J. Challinor Title Page

1Met Oﬃce Hadley Centre, FitzRoy Road, Exeter, EX1 3PB, UK Abstract Introduction 2School of Earth and Environment, Institute for Climate and Atmospheric Science, Conclusions References University of Leeds, Leeds, LS2 9AT, UK Tables Figures Received: 16 April 2015 – Accepted: 29 April 2015 – Published: 16 June 2015 Correspondence to: C. Mathison (camilla.mathison@metoﬃce.gov.uk) J I Published by Copernicus Publications on behalf of the European Geosciences Union. J I

Back Close

Full Screen / Esc

Printer-friendly Version

Interactive Discussion

5789 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion Abstract HESSD South Asia is a region with a large and rising population and a high dependance on industries sensitive to water resource such as agriculture. The climate is hugely 12, 5789–5840, 2015 variable with the region relying on both the Asian Summer Monsoon (ASM) and glaciers 5 for its supply of fresh water. In recent years, changes in the ASM, fears over the rapid South Asia river flow retreat of glaciers and the increasing demand for water resources for domestic and projections and their industrial use, have caused concern over the reliability of water resources both in the implications for water present day and future for this region. The climate of South Asia means it is one of the resources most irrigated agricultural regions in the world, therefore pressures on water resource 10 affecting the availability of water for irrigation could adversely affect crop yields and C. Mathison et al. therefore food production. In this paper we present the first 25 km resolution regional climate projections of river flow for the South Asia region. ERA-Interim, together with two global climate models (GCMs), which represent the present day processes, Title Page particularly the monsoon, reasonably well are downscaled using a regional climate Abstract Introduction 15 model (RCM) for the periods; 1990–2006 for ERA-Interim and 1960–2100 for the two GCMs. The RCM river flow is routed using a river-routing model to allow analysis of Conclusions References present day and future river flows through comparison with river gauge observations, Tables Figures where available.

In this analysis we compare the river ﬂow rate for 12 gauges selected to represent J I 20 the largest river basins for this region; Ganges, Indus and Brahmaputra basins and characterize the changing conditions from east to west across the Himalayan J I arc. Observations of precipitation and runoﬀ in this region have large or unknown Back Close uncertainties, are short in length or are outside the simulation period, hindering model development and validation designed to improve understanding of the water cycle for Full Screen / Esc 25 this region. In the absence of robust observations for South Asia, a downscaled ERA- Interim RCM simulation provides a benchmark for comparison against the downscaled Printer-friendly Version GCMs. On the basis that these simulations are among the highest resolution climate Interactive Discussion simulations available we examine how useful they are for understanding the changes

5790 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion in water resources for the South Asia region. In general the downscaled GCMs capture the seasonality of the river flows, with timing of maximum river flows broadly matching HESSD the available observations and the downscaled ERA-Interim simulation. Typically the 12, 5789–5840, 2015 RCM simulations over-estimate the maximum river flows compared to the observations 5 probably due to a positive rainfall bias and a lack of abstraction in the model although comparison with the downscaled ERA-Interim simulation is more mixed with only South Asia river flow a couple of the gauges showing a bias compared with the downscaled GCM runs. projections and their The simulations suggest an increasing trend in annual mean river flows for some of implications for water the river gauges in this analysis, in some cases almost doubling by the end of the resources 10 century; this trend is generally masked by the large annual variability of river flows for this region. The future seasonality of river flows does not change with the future C. Mathison et al. maximum river flow rates still occuring during the ASM period, with a magnitude in some cases, greater than the present day natural variability. Increases in river flow Title Page during peak flow periods means additional water resource for irrigation, the largest 15 usage of water in this region, but also has implications in terms of inundation risk. Abstract Introduction Low flow rates also increase which is likely to be important at times of the year when Conclusions References water is historically more scarce. However these projected increases in resource from rivers could be more than countered by changes in demand due to reductions in the Tables Figures quantity and quality of water available from groundwater, increases in domestic use 20 due to a rising population or expansion of other industries such as hydro-electric power J I generation. J I

Back Close 1 Introduction Full Screen / Esc South Asia, the Indo-Gangetic plain in particular, is a region of rapid socio-economic change where both population growth and climate change is expected to have a large Printer-friendly Version 25 impact on available water resource and food security. The region is home to almost Interactive Discussion 1.6 billion people and the population is forecast to increase to more than 2 billion by 2050 (United Nations, 2013). The economy of this region is rural and highly 5791 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion dependant on climate sensitive sectors such as the agricultural and horticultural industry, characterised by a large demand for water resources. As a result, over the HESSD coming decades, the demand for water from all sectors; domestic, agricultural and 12, 5789–5840, 2015 industrial is likely to increase (Gupta and Deshpande, 2004; Kumar et al., 2005). 5 The climate of South Asia is dominated by the Asian Summer Monsoon (ASM), with much of the water resource across the region provided by this climatological South Asia river flow phenomena during the months of June–September (Goswami and Xavier, 2005). The projections and their contribution from glacial melt to water resources is less certain but likely to be important implications for water outside the ASM period during periods of low river flow (Mathison et al., 2013). Glaciers resources 10 and seasonal snowpacks are natural hydrological buffers releasing water during spring and autumn when the flows of catchments like the Ganges are at their lowest. Similarly C. Mathison et al. they may act to buffer inter-annual variability as well releasing water during warmer drier years and accumulating during wetter colder years (Barnett et al., 2005). Recent Title Page studies have shown that both of these are changing (ASM rainfall – Christensen et al., 15 2007, and glacier mass balance – Fujita and Nuimura, 2011) putting more pressure Abstract Introduction on groundwater resources which is not sustainable in the longer term (Rodell et al., Conclusions References 2009). Gregory et al.(2005) suggest that the availability and quality of ground water for irrigation could be more important factors influencing food security than the direct Tables Figures effects of climate change, particularly for India. Aggarwal et al.(2012) suggest that an 20 increase in extremes (both temperature and precipitation) could lead to instability in J I food production and it is this variability in food production that is potentially the most significant effect of climate change for the South Asia region. J I Immerzeel et al.(2010) found that by the 2050s the main upstream water supply Back Close could decrease by approximately 18 % although this decrease was partly offset by an Full Screen / Esc 25 8 % increase in precipitation. Immerzeel et al.(2010) use general circulation models (GCMs) which have a coarse resolution and are known to have difficulty in capturing Printer-friendly Version the monsoon precipitation and in estimating the relationship between daily mean temperature and melting of snow and ice. Interactive Discussion

5792 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion The Indo-Gangetic plains have traditionally provided the staple crops of rice and wheat (Aggarwal et al., 2000) for India and South Asia as a whole, irrigation is an HESSD important part of this industry and any limitation of water resource needed to maintain 12, 5789–5840, 2015 yields of these crops could have implications on the food and water security of the 5 region. The aim of this analysis is to examine how useful these simulations are for understanding how river flows could change in South Asia in the future and the South Asia river flow implications this could have on water resources that are increasingly in demand. The projections and their water resources for the South Asia region as a whole are generally poorly understood implications for water with limitations in the observing networks and availability of data for both precipitation resources 10 and river flows presenting a real challenge for validating models and estimates of the water balance of the region. In this analysis we use a 25 km resolution regional C. Mathison et al. climate model (RCM) with a demonstrated ability to capture the ASM to downscale ERA-interim re-analysis data (Simmons et al., 2007) and two GCMs able to capture Title Page the main features of the large-scale circulation (Annamalai et al., 2007; Mathison et al., 15 2013). In the absence of robust observations, particularly for high elevation regions like Abstract Introduction the Himalaya, the ERA-interim simulation provides a constrained estimate of the water Conclusions References balance of the region. In a previous study, Akhtar et al.(2008) found that RCM data produced better results when used with a hydrological model than using poor-quality Tables Figures observation data; this implies greater confidence in the RCM simulated meteorology 20 than available observational data for this region (Wiltshire, 2013). Therefore in this J I analysis, as in Wiltshire(2013), in addition to the observations that are available, it is appropriate to use the ERA-interim simulation as a benchmark against which to J I evaluate the GCM driven regional simulations. The RCM includes a land-surface model Back Close which includes a full physical energy-balance snow model (Lucas-Picher et al., 2011) Full Screen / Esc 25 providing an estimate of the gridbox runoff which is then used to drive the Total Runoff Integrating Pathways river routing model (TRIP; Oki and Sud, 1998) in order to present Printer-friendly Version 25 km resolution regional climate projections of riverflow for the South Asia region. TRIP has been used previously in Falloon et al.(2011) which used GCM outputs Interactive Discussion directly to assess the skill of a global river-routing scheme. TRIP is applied here to

5793 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion runoff from a subset of the 25 km resolution RCM simulations completed as part of the EU HighNoon project (HNRCM) to provide river flow rates for South Asia. A selection of HESSD river flow gauges, mainly from the GRDC (GRDC, 2014) network provide observations 12, 5789–5840, 2015 which are used, in addition to downscaled ERA-interim river flows, to evaluate the 5 downscaled GCM river flows for the major catchments of the South Asia region; these river gauges aim to illustrate from the perspective of river flows as modelled in an South Asia river flow RCM, that the influence of the ASM on precipitation totals increases, from west to projections and their east and north to south across the Himalayan mountain range, while that of western implications for water disturbances reduces (Wiltshire, 2013; Dimri et al., 2013; Ridley et al., 2013; Collins resources 10 et al., 2013). The differing influences across the Himalayan arc result in complex regional differences in sensitivity to climate change; with western regions dominated C. Mathison et al. by non-monsoonal winter precipitation and therefore potentially less susceptible to reductions in annual snowfall (Wiltshire, 2013; Kapnick et al., 2014). The selection of Title Page these gauges and the models used are described in Sect.2, while a brief evaluation of 15 the driving data and the river flow analysis is presented in Sect.3. The implications of Abstract Introduction the potential changes in river flows on water resources and conclusions are discussed Conclusions References in Sects.4 and5 respectively. Tables Figures

2 Methodology J I

2.1 Observations J I

Back Close 20 The total precipitation within each of the downscaled GCM simulations are compared against a downscaled ERAinterim simulation and precipitation observations from Full Screen / Esc the Asian Precipitation-Highly Resolved Observational Data Integration Towards the

Evaluation of Water Resources (APHRODITE – Yatagai et al., 2012) dataset in Printer-friendly Version Sect. 3.1 focusing on the main river basins in the region and included in the river Interactive Discussion 25 flow analysis (in Sect. 3.2); the Indus and the Ganges/Brahmaputra. The precipitation patterns for each basin are useful for understanding the changes in the river 5794 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion flows within the catchments, however, rain gauges in the APHRODITE dataset are particularly sparse at higher elevations (see Yatagai et al., 2012, Fig. 1) which leads to HESSD underestimation of the basin wide water budgets particularly for mountainous regions 12, 5789–5840, 2015 (Andermann et al., 2011). Therefore the reanalysis product ERAinterim (Simmons 5 et al., 2007) is also used as a benchmark to compare the downscaled GCMs against. All of the gauges selected for the river flow analysis presented here lie within these South Asia river flow river catchments and are chosen to characterize the conditions along the Himalayan projections and their arc using river flow data from the Global Runoff Data Centre (GRDC, 2014). A brief implications for water geographical description of the rivers and the chosen gauges is given in this section, resources 10 their locations are shown in Fig.1 and listed in Table1 (including the abbreviations shown in Fig.1 and the gauge location in terms of latitude and longitude). C. Mathison et al. The Indus, originates at an elevation of more than 5000 m in western Tibet on the northern slopes of the Himalayas, flowing through the mountainous regions of India Title Page and Pakistan to the west of the Himalayas. The upper part of the Indus basin is greatly 15 influenced by western disturbances which contribute late winter snowfall to the largest Abstract Introduction glaciers and snow fields outside the polar regions; the meltwaters from these have Conclusions References a crucial role in defining the water resource of the Indus basin (Wescoat Jr., 1991). In this analysis the Attock gauge is the furthest upstream and the Kotri gauge, located Tables Figures further downstream provide observations on the main trunk of the Indus river. The 20 Chenab river, located in the Panjnad basin and in this analysis represented by the J I Panjnad gauge, is a major eastern tributary of the Indus, originating in the Indian state of Himachal Pardesh. In the upper parts of the Chenab sub-basin western disturbances J I contribute considerably to precipitation while the foothills are also influenced by the Back Close ASM (Wescoat Jr., 1991). Full Screen / Esc 25 The Ganges river originates on southern slopes of the Himalayas (Thenkabail et al., 2005) and traverses thousands of kilometres before joining with the Brahmaputra Printer-friendly Version in Bangladesh and emptying into the Bay of Bengal (Mirza et al., 1998). The Ganges basin has a population density 10 times the global average making it the Interactive Discussion most populated river basin in the world (Johnston and Smakhtin, 2014), it covers

5795 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion 1.09 millionkm2 with 79 % in India, 13 % in Nepal, 4 % in Bangladesh and 4 % in China (Harding et al., 2013). The main trunk of the Ganges is represented in this analysis by HESSD the gauge at the Farakka barrage, located at the India–Bangladeshi border, to the East 12, 5789–5840, 2015 of the Himalayas. The Bhagirathi river, located in the region often referred to as the 5 Upper Ganga basin, is one of the main head streams of the Ganges. The Bhagirathi river originates from Gaumukh 3920 ma.s.l. at the terminus of the Gangotri glacier in South Asia river ﬂow Uttarakhand, India (Bajracharya and Shrestha, 2011). The Tehri dam is located on this projections and their tributary, providing the most central data point on the Himalayan arc in this analysis implications for water (this is not a GRDC gauge). resources 10 The Karnali river (also known as Ghaghara), drains from the Himalaya originating in C. Mathison et al. Nepal ﬂowing across the border to India where it drains into the Ganges. The Karnali is the largest river in Nepal and a major tributary of the Ganges (Bajracharya and

Shrestha, 2011) accounting for approximately 11 % of the Ganges discharge, 5 % of Title Page its area and 12 % of its snowfall in the HNRCMs. Two of the river gauges in this Abstract Introduction 15 analysis; the Benighat and the Chisapani are located on this river. Two other sub-

catchments complete those covering the Ganges basin; the Narayani river (also known Conclusions References as the Gandaki River, represented here by the Devghat river gauge); reportedly very dependant on glaciers at low flow times of the year with over 1700 glaciers covering Tables Figures more than 2200 km2 (Bajracharya and Shrestha, 2011). The Arun river, part of the 20 Koshi river basin originates in Tibet, flows south through the Himalayas to Nepal. The J I Arun, represented in this analysis by the Turkeghat gauge joins the Koshi river which J I flows in a southwest direction as a tributary of the Ganges. The Brahmaputra originates from the glaciers of Mount Kailash at more than Back Close 5000 ma.s.l., on the northern side of the Himalayas in Tibet flowing into India, and Full Screen / Esc 25 Bangladesh before merging with the Padma in the Ganges Delta. The Brahmaputra

is prone to flooding due to its surrounding orography and the amount of rainfall the Printer-friendly Version catchment receives (Dhar and Nandargi, 2000). The Brahmaputra is represented in this analysis by three gauges; Yangcun, the highest upstream gauge, Pandas in the Interactive Discussion middle and Bahadurabad furthest downstream but above the merge with the Padma. 5796 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion 2.2 Models HESSD This analysis utilizes 25 km resolution regional climate modelling of the Indian subcontinent to provide simulations across the Hindu-Kush Karakoram Himalaya mountain 12, 5789–5840, 2015 belt. To sample climate uncertainty, two GCM simulations that have been shown to 5 capture a range of temperatures and variability in precipitation similar to the AR4 South Asia river flow ensemble for Asia (Christensen et al., 2007) and that have been shown to simulate projections and their the ASM (Kumar et al., 2013; Annamalai et al., 2007); The Third version of the Met implications for water Office Hadley Centre Climate Model (HadCM3 – Pope et al., 2000; Gordon et al., resources 2000, a version of the Met Office Unified Model) and ECHAM5 (3rd realization – 10 Roeckner et al., 2003) are downscaled using the HadRM3 (Jones et al., 2004) RCM. C. Mathison et al. An ERA-interim (Simmons et al., 2007) driven RCM simulation is also shown to provide a benchmark for comparison against the GCM driven simulations in the absence of good quality observations (see Sects. 2.1 and 3.1). The RCM simulations are Title Page performed at 25 km, part of the ensemble produced for the EU-HighNoon program, Abstract Introduction ◦ ◦ ◦ ◦ 15 for the whole of the Indian subcontinent (25 N, 79 E–32 N, 88 E) and are currently the finest resolution modelling available for this region (Mathison et al., 2013; Moors Conclusions References et al., 2011; Kumar et al., 2013). There are 19 atmospheric levels and the lateral Tables Figures atmospheric boundary conditions are updated 3 hourly and interpolated to a 150 s

timestep. The experimental design of the HighNoon ensemble compromises between J I 20 the need for higher resolution climate information for the region, the need for a number of ensemble members to provide a range of uncertainty and the limited number of J I GCMs that are able to simulate the ASM. These factors are all important given the Back Close limited computational resources available. In these simulations the land surface is represented by version 2.2 of the Met Full Screen / Esc 25 Office Surface Exchange Scheme (MOSESv2.2, Essery et al., 2003). MOSESv2.2 treats subgrid land-cover heterogeneity explicitly with separate surface temperatures, Printer-friendly Version radiative fluxes (long wave and shortwave), heat fluxes (sensible, latent and ground), Interactive Discussion canopy moisture contents, snow masses and snowmelt rates computed for each

5797 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion surface type in a grid box (Essery et al., 2001). However the air temperature, humidity and wind speed above the surface are treated as homogenous across the gridbox HESSD and precipitation is applied uniformly over the different surface types of each gridbox. 12, 5789–5840, 2015 The relationship between the precipitation and the generation of runoff is complicated, 5 depending on not only the intensity, duration and distribution of the rainfall but also the characteristics of the surface e.g. the infiltration capacity of the soil, the vegetation South Asia river flow cover, steepness of the orography within the catchment and the size of the catchment projections and their (Linsley et al., 1982). In GCMs and even 25 km RCMS such as the ones presented implications for water here, the resolution is often too coarse to explicitly model the large variations of resources 10 soil moisture and runoff within a catchment and therefore the major processes are parameterized (Gedney and Cox, 2003). The method used within MOSES2.2 for C. Mathison et al. generating surface and subsurface runoff across a gridbox is through partitioning the precipitation into interception by vegetation canopies, throughfall, runoff and infiltration Title Page for each surface type (Essery et al., 2003). The Dolman and Gregory(1992) infiltration 15 excess mechanism generates surface runoff; this assumes an exponential distribution Abstract Introduction of point rainfall rate across the fraction of the catchment where it is raining (Clark and Conclusions References Gedney, 2008). Moisture fluxes are allowed between soil layers; these are calculated using the Darcy equation, with the water going into the top layer defined by the gridbox Tables Figures average and any excess removed by lateral flow (Essery et al., 2001). Excess moisture 20 in the bottom soil layer drains from the bottom of the soil column at a rate equal to J I the hydraulic conductivity of the bottom layer as subsurface runoff (Clark and Gedney, 2008). The performance of MOSESv2.2 is discussed in the context of a GCM in Essery J I et al.(2001), however no formal assessment of MOSESv2.2 and the runo ff generation Back Close in particular has been done for the RCM. Full Screen / Esc 25 In this analysis the simulated runoff is converted into river flow using the TRIP river routing scheme (Oki and Sud, 1998) as a post-processing step. TRIP is a simple model Printer-friendly Version that moves water along a pre-defined 0.5◦ river network; the Simulated Topological Network at 30 min resolution (STN-30p, version 6.01; Vörösmarty et al., 2000a,b; Interactive Discussion Fekete et al., 2001) in order to provide mean runoff per unit area of the basin which can

5798 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion be compared directly with river gauge observations. The TRIP model has been shown to agree well with observed river flow gauge data (Oki et al., 1999) and largely showed HESSD good skill when comparing run off from several land surface models (Morse et al., 12, 5789–5840, 2015 2009). Implementation of TRIP in two GCMs; HadCM3 and HadGEM1 is described 5 by Falloon et al.(2007) and was found to improve the seasonality of the river flows into the ocean for most of the major rivers. Using TRIP ensures the river flow forcing South Asia river flow is consistent with the atmospheric forcing, however it also assumes that all runoff is projections and their routed to the river network and as such there is no net aquifer recharge/discharge. implications for water This may not be the case in regions with significant ground water extraction which resources 10 is subsequently lost though evaporation and transported out of the basin. These simulations do not include extraction, which for this region is large, particularly for C. Mathison et al. irrigation purposes (Biemans et al., 2013); this means that the extraction-evaporation and subsequent recycling of water in a catchment (Harding et al., 2013; Tuinenburg Title Page et al., 2014) is not considered in this analysis. The routed runoff of the HNRCM 15 simulations are referred to here using only the global driving data abbreviations; Abstract Introduction ERAint, ECHAM5 and HadCM3. Conclusions References

Tables Figures 3 Results

3.1 Comparison of present day driving data with observations J I J I In this section we summarise the main points from previous analysis and evaluation Back Close 20 of the HNRCM simulations that provide the driving data for the river ﬂow projections

(Kumar et al., 2013; Lucas-Picher et al., 2011; Mathison et al., 2013). We also look Full Screen / Esc again at the total precipitation for these simulations focussing on the major river basins

for the region before presenting the river flow projections for individual gauges in Printer-friendly Version Sect. 3.2. Lucas-Picher et al.(2011) evaluates the ability of RCMs to capture the ASM Interactive Discussion 25 using ERA-40 data, Kumar et al.(2013), as part of the HighNoon project, completes analysis using the HNRCMs forced with ERA-Interim data. The HNRCM simulations 5799 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion are themselves evaluated against a range of observations for the Ganges/Brahmaputra river basin in Mathison et al.(2013). Figure2 shows the spatial distribution of total HESSD precipitation for the monsoon period (June to September; Goswami and Xavier, 2005) 12, 5789–5840, 2015 for APHRODITE observations together with the downscaled ERAint and GCM driven 5 simulations. Figure2 highlights that, in general the HNRCM simulations capture the spatial characteristics of the ASM, successfully reproducing regions of high convective South Asia river flow precipitation, maximum land rainfall and the rain shadow over the east coast of India as projections and their described in more detail in Kumar et al.(2013). The RCMs are also able to reproduce implications for water the inter-annual variability of the region although they underestimate the magnitude of resources 10 the variation (Kumar et al., 2013). In general the GCMs in the AR4 ensemble exhibit cold and wet biases compared to observations both globally (Nohara et al., 2006) and C. Mathison et al. for South Asia (Christensen et al., 2007), although these are generally reduced in the RCM simulations there is a cold bias in the RCM that is probably carried over from the Title Page larger bias in the GCMs (Mathison et al., 2013; Kumar et al., 2013). 15 Figures3 and4 show the annual mean and the monthly climatology of Abstract Introduction the total precipitation for the RCM simulations, compared with 25 km resolution Conclusions References APHRODITE observations, for the main basins in this analysis; the Indus and the Ganges/Brahmaputra. The Ganges and Brahmaputra catchments are considered Tables Figures together in this analysis as these rivers join together in the Ganges Delta and within 20 TRIP there is no clear delineation between the two catchments. In general the models J I appear to over estimate the seasonal cycle of total precipitation (Fig.4) compared with the APHRODITE observations; this is highlighted by the annual mean of the J I total precipitation shown in Fig.3. However, the sparsity of the observations at high Back Close elevations dicussed in Sect. 2.1 make it difficult to attach error bars to the observations Full Screen / Esc 25 particularly for mountainous regions and therefore an ERAint simulation is used to provide a benchmark for comparison against the two downscaled GCM simulations. Printer-friendly Version The annual mean (Fig.3) and the monthly climatology (Fig.4) show that, for these catchments, the ERAint simulation lies between the two HighNoon ensemble members Interactive Discussion

5800 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion except during peak periods of precipitation when the magnitude of the total precipitation in the ERAint simulation is larger. HESSD The seasonal cycles of total precipitation are distinctly different between the basins 12, 5789–5840, 2015 shown. The Indus basin (Fig.4, left), indicates two periods of precipitation; one 5 smaller peak between January and May and another larger one between July and September. The smaller peak occurs later than both ERAint and the observations South Asia river flow for the downscaled GCM simulations while the timing of the larger peak compare projections and their well between the observations, ERAint and the downscaled GCM simulations. implications for water The magnitude of the peaks in precipitation in the APHRODITE observations are resources 10 consistently lower throughout the year than the simulations. The magnitude of the ERAint total precipitation is typically larger than both GCM driven simulations while C. Mathison et al. the ECHAM5 simulation is the lowest and closest to the APHRODITE observations, HadCM3 is between ECHAM5 and ERAint for most of the year. In contrast the Title Page Ganges/Brahmaputra catchment (Fig.4, right) has one strong peak between July and 15 September; this cycle is also captured reasonably well by the simulations, both in Abstract Introduction terms of magnitude and timing of the highest period of precipitation. However there Conclusions References is a tendancy for the simulations to overestimate rainfall between January and June compared to the observations, thus lengthening the wet season (Mathison et al., 2013). Tables Figures Mathison et al.(2013) also show that in these simulations, the region of maximum 20 precipitation along the Himalayan foothills is displaced slightly to the north of that shown J I in the observations. One explanation for this could be that the peak in total precipitation is due the distribution of observations already discussed. Alternatively it could be due J I to the model resolution, which may, at 25 km still be too coarse to adequately capture Back Close the influence of the orography on the region of maximum precipitation and therefore it Full Screen / Esc 25 is displaced from where it actually occurs. The downscaled ERAint simulation also indicates a higher total precipitation for January–May that is within the range of Printer-friendly Version uncertainty of the GCM driven simulations. However for the remainder of the Monsoon period, ERAint has a higher total precipitation than the GCM driven simulations; this is Interactive Discussion highlighted by the spatial distribution of total precipitation shown in Fig.2 which shows

3.2 Present day modelled river ﬂows 12, 5789–5840, 2015

In this section we compare present day modelled river flows with observations and South Asia river flow 5 a downscaled ERAint simulation using annual average river flows (see Fig.5) and projections and their monthly climatologies (see Fig.6). implications for water The annual average river flow rates for each river gauge (described in Sect. 2.1) are resources shown by the paler lines in Fig.5 (red line-HadCM3, blue line-ECHAM5) with the darker lines showing a smoothed average to highlight any visible trends in the simulations. C. Mathison et al. 10 The plots show the model data for the whole period of the simulations including the historical period for each of the simulations and the available observations (GRDC, 2014 – black line) for that location. It is clear from this plot that observed river flow Title Page data is generally limited which makes statistical analysis of the observations difficult. Abstract Introduction River flow data for this region is considered sensitive and is therefore not readily 15 available particularly for the present day. For each of the gauges shown here, there Conclusions References are generally several complete years of data but often the time the data was collected Tables Figures pre-dates the start of the model run. The ERAint simulation is also shown (cyan line- ERAint) to provide a benchmark in the absence of well-constrained observations (See J I Sect. 3.1). The comparison between the model and observations shown in Figs.5 and 20 6 is therefore to establish if the model and observations are comparable in terms of J I

the average seasonal cycle and mean river ﬂow rate without over-interpreting how well Back Close they replicate the observations. The multi-year monthly mean modelled river ﬂows for ECHAM5 (blue line), HadCM3 Full Screen / Esc (red line), for the period 1971–2000 and ERAint (cyan line) for the period 1990– 25 2007 are shown for each river gauge location in Fig.6. The multi-year mean for all Printer-friendly Version

the available observations are also shown (Fig.6, black line – GRDC(2014) except Interactive Discussion for the Tehri Dam on the Bhagirathi river for which observations are not shown but were received via personal communication from the Tehri Dam operator). The shaded 5802 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion regions show the 1.5 standard deviation (SD) from the mean for each GCM driven model for the 1971–2000 period which represents the variability of the region and HESSD provides a plausible range of river flows in the absence of any known observation 12, 5789–5840, 2015 errors for the GRDC observations (personal communication, GRDC). Estimates of 5 observation errors for river gauges vary in the literature with a recommendation in Falloon et al.(2011) for GCMs to be consistently within 20 % of the observations while South Asia river flow Oki et al.(1999) suggest that errors of 5 % at the 95 % confidence interval might be projections and their expected. McMillan et al.(2010) propose a method for quantifying the uncertainty implications for water in river discharge measurements by defining confidence bounds. Therefore in this resources 10 analysis, where the 1.5 SD range encompasses the observations and ERAint, given the variability of the region and the limitations of the observations, this is considered C. Mathison et al. a reasonable approximation. The Kotri gauge on the Indus (Fig.6, 1st row, left column) and the Yangcun gauge Title Page on the Brahmaputra (Fig.6, 6th row, left column) are the only two gauges where 15 the modelled river flow is higher than the observations and not within the estimated Abstract Introduction variability (1.5 SD) of the region. The ERAint simulation is also outside the estimated Conclusions References variability (1.5 SD) for the Benighat gauge on the Karnali river (Fig.6, 3rd row, left column). The differences in these gauges are also reflected in the annual mean river Tables Figures flows (Fig.5) for these river gauges which are higher than observed. The explanation 20 for the river flow at the Kotri gauge being too high could be due to the extraction of J I water which is not included in the model; this is particularly plausible for this gauge as this is a downstream gauge located relatively close to the river mouth and the Indus J I has a relatively large extraction rate (Biemans et al., 2013). The Yangcun gauge is Back Close a more upstream gauge and the differences between the model and observations for Full Screen / Esc 25 this gauge are more likely to be related to the precipitation in the simulations which is high at this location, particularly during the ASM (see Fig.2); this could be having Printer-friendly Version a direct effect on the riverflow. At the other two gauges on the Brahmaputra downstream of the Yangcun gauge; Interactive Discussion the Pandu and Bahadurabad (Fig.6, 6th row, right column and 5th row, right column

5803 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion respectively), the seasonal cycle of river flow has a very broad peak particularly in the modelled river flows compared to the other gauges. In the simulations the snowfall HESSD climatology for the Ganges/Brahmaputra basin (not shown) has a similar seasonal 12, 5789–5840, 2015 cycle to that of the river flow for the Bahadurabad and the Pandu gauges. It is therefore 5 likely that the broad peak in river flow is related to the broad peak in snowfall and subsequent snowmelt. The Pandu gauge is also one of only two gauges where the South Asia river flow modelled river flow is less than the observations for at least part of the year, the other projections and their being the Devghat gauge on the Narayani river (Fig.6, 4th row, left column); both of implications for water these gauges are located in the Himalayan foothills close to the region of simulated resources 10 maximum total precipitation. If the simulations put the location of this maximum below these gauges this could cause the river flows at the gauges to be lower than observed. C. Mathison et al. The river flow on the main trunk of the Ganges at the Farakka barrage (shown in Fig.6, 5th row, left column), is a reasonable approximation to the observations in terms of Title Page magnitude, however the timing of the peak flow seems to be later in the models. It could 15 be argued this also happens in some of the other gauges although it is more noticeable Abstract Introduction for the Farakka barrage. All the gauges shown here are for glacierized river basins Conclusions References and although snow fields and therefore snow melt are represented and the models will replicate some aspects of melt affecting river flow, glacial melt is not explicitly Tables Figures represented in the RCM used for these simulations; this could be important for the 20 timing and magnitude of the maximum and minimum river flows for these catchments. J I

3.3 Future river flows J I Back Close This section considers the future simulations from the RCM in terms of both precipitation and river flows to establish any implications for future water resources. Full Screen / Esc The future annual means of both total precipitation (for the two main basins covering 25 the gauges in this analysis) and river flows (for each gauge) are shown in Figs.3 Printer-friendly Version

and5 respectively. In both Figs.3 and5 the annual average is shown for the two model Interactive Discussion simulations (red line-HadCM3 and blue line-ECHAM5) by the unsmoothed (paler) lines; the smoothed (darker) lines aim to highlight any trends in the data that might be masked 5804 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion by the high variability shown in the annual mean of the future projections of both precipitation and river flow. HESSD Figure3 also highlights the variability in the future projections of total precipitation for 12, 5789–5840, 2015 South Asia between basins; in these simulations the Ganges/Brahmaputra catchment 5 shows an increasing trend in total precipitation and more variation between the simulations (Fig.3, right) than the Indus basin (Fig.3, left), which has a much flatter South Asia river flow trajectory to 2100. projections and their The trends shown by the smoothed (darker) lines overlaid on top of the annual implications for water mean river flows shown in Fig.5 highlight an upward trend in river flows at some of resources 10 the gauges, in particular, the Narayani-Devghat (Fig.5, 4th row, left column), Arun- Turkeghat (Fig.5, 4th row, right column) and Ganges-Farakka (Fig.5, 5th row, left C. Mathison et al. column) all show an upward trend toward the 2100s that actually represents a doubling of the river flow rate which could be important for water resources for the region. Title Page In the following analysis in Sects. 3.3.1 and 3.3.3 we focus on the modelled river flow 15 for two future periods; 2040–2070 (referred to as 2050s) and 2068–2098 (referred to Abstract Introduction as 2080s). In Sect. 3.3.1 we consider the mean seasonal river flow for the two periods, Conclusions References to establish if there are changes in the seasonality of river flows in the future before focussing on the upper and lower 10 % of the river flows for the two future periods in Tables Figures Sect. 3.3.3. Section 3.3.4 continues to focus on the highest and lowest flows but uses 20 the 10th and 90th percentile for each decade to compare models for each gauge. J I

3.3.1 Climatology analysis J I Back Close The seasonal cycle of modelled river ﬂows at each of the river gauge locations are shown in Fig.7 for two future periods; 2050s (solid lines) and 2080s (dashed lines) Full Screen / Esc for the two ensemble members (HadCM3 – red lines, ECHAM5 – blue lines). The 25 shaded part of the plot represents the present day natural variability using the 1.5 Printer-friendly Version

SD of the 1971–2000 period from each model. South Asia is a very variable region, Interactive Discussion yet these models suggest the future mean river flow could lie outside the present day variability for peak flows for some of the gauges in this study; this could have important 5805 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion implications for water resources for the region. The gauges that show an increase in maximum river flows in Fig.7 are mainly those in the middle of the Himalayan HESSD arc as shown in Fig.1 with the western most (Indus gauges) and the eastern most 12, 5789–5840, 2015 (Brahmaputra gauges) typically still within the range of present day variability. This 5 could be due to the changes in the influence on river flow from west to east becoming more influenced by the ASM and less by western disturbances, with basins in the South Asia river flow centre of the Himalayas and to the north influenced by both phenomena. Figure7 also projections and their suggests that the maximum river flows still occur mainly during the ASM for many of implications for water the gauges shown. As mentioned in Sect. 3.2 glacial melt is not explicitly represented resources 10 in the RCM used for these simulations and this could have implications for the timing and magnitude of the future high and low river flows for these catchments. C. Mathison et al. Analysis of the 30 year mean is useful for understanding the general climatology of the region but often the mean does not provide the complete picture particularly Title Page when it is the periods of high and low river flow that are critical in terms of water 15 resources. Mathison et al.(2013) highlight the importance of potential changes in the Abstract Introduction seasonal maximum and minimum river flows for the agricultural sector. The analysis in Conclusions References Sect. 3.3.2 considers the distribution of river flows across the region using the same river gauges and also considers changes in the upper and lower parts of the distribution Tables Figures of river flow. J I 20 3.3.2 High and low flow analysis J I

The distributions of the river flows for each of the gauges are shown in the form of Back Close probability density functions (pdfs), calculated using Kernal Density Estimation (KDE, Scott, 2009; Silverman, 1986) in Fig.8. Figure8 illustrates how the lowest flows Full Screen / Esc dominate the distribution. In most of the gauges and both models the 1971–2000 25 period has the highest frequency of the lowest flows, the curves then tend to flatten Printer-friendly Version

in the middle of the distribution before tailing oﬀ toward zero for the frequency of the Interactive Discussion highest ﬂows.

5806 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion The Yangcun gauge on the Brahmaputra (Fig.8, 6th row, left column) shows the least change of all the gauges between the 1971–2000 period, future periods and models, HESSD however the the distributions for the gauges downstream of Yangcun; the Pandu (Fig.8, 12, 5789–5840, 2015 6th row, right column) and the Bahadurabad (Fig.8, 5th row, right column) are notable 5 for their differences. The Pandu and Bahadurabad gauges have two distinct peaks in frequency, one toward the lower end of the river flow distribution, consistent with the South Asia river flow other gauges shown, and another in the middle of the distribution, where the distribution projections and their for most other gauges flattens out. This is consistent with the broader peak in the implications for water seasonal cycle shown for these gauges in Fig.7 and could be explained by snowmelt resources 10 (see Sect. 3.2). In some of the other gauges this peak in the middle of the range of river flows is evident to a much lesser degree but tends to be restricted to the two C. Mathison et al. future periods and is not evident in the present day distribution e.g. the two Karnali river gauges (Fig.8, 3rd row). For the two future periods there is a similar shape to the Title Page distributions for each of the river gauges compared to the 1971–2000, however there 15 is a tendancy for a reduction in the frequency of the lowest flows and an increase in Abstract Introduction the magnitude of the highest flows for both models across the gauges. Conclusions References In the analysis that follows, the changes in the lowest and highest 10 % of flows are considered in more detail using two alternative approaches; one comparing the 10th Tables Figures and 90th percentile for each model for each decade and the other takes the relevant 20 percentiles for the 1971–2000 period and uses these as thresholds for the two future J I periods. J I

3.3.3 Threshold analysis Back Close

In the pdfs shown in Fig.8, the individual distributions for the gauges shown suggest Full Screen / Esc that the occurrance of the lowest flows is reducing and the magnitudes of the higher 25 flows are increasing toward the end of the century. This analysis aims to confirm this Printer-friendly Version

pattern by comparing the two future periods (2050s and 2080s) against the 1971– Interactive Discussion 2000 period explicitly using thresholds defined by the 10th and 90th percentiles for this present day period for each river gauge. Graphical examples from the results of 5807 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion this analysis are shown for the Farakka Barrage on the River Ganges in Fig.9, which shows the number of times river flows are less than the (1971–2000) 10th percentile HESSD and Fig. 10, which shows the number of times river flows are greater than the (1971– 12, 5789–5840, 2015 2000) 90th percentile. Each of the plots in Figs.9 and 10 show a di fferent decade; 5 historical (top), 2050s (middle) and 2080s (bottom). In Fig.9 the number of times the model is below the 1971–2000 threshold reduces in each of the future decades South Asia river flow and in Fig. 10 the number of points increases in each of the future decades. Table2 projections and their summarises the main results for each of the gauges from this analysis by providing the implications for water percentage change in the number of times the model simulations is less than the 10th resources 10 or greater than the 90th percentile for the 1971–2000 thresholds. Table2 illustrates that the patterns shown in Figs.9 and 10 are generally true for almost every other C. Mathison et al. gauge in the analysis. The Tehri Dam (Bhagirathi) is the only exception of the gauges shown in Table2, showing an increase of 12 % in the number of incidences where the Title Page river flow is less than the 1971–2000 10th percentile for the 2080s; this is mainly due 15 to the ECHAM5 model which has a high number of incidences. The Yangcun gauge Abstract Introduction (Brahmaputra) is the only gauge where there is no change in the number of incidences Conclusions References where the river flow is less than the 10th percentile for 1971–2000 in either the 2050s or the 2080s, probably because the lowest river flows are already very low at this gauge. Tables Figures In every gauge there is an increase in the number of incidences where river flows are 20 greater than the 90th percentile for 1971–2000 in the 2050s and 2080s in these model J I runs, with several of the gauges suggesting increases in the number of events above the 90th percentile for the 1971–2000 period of more than 100 %. This confirms the J I conclusions drawn visually from Fig.8 that both low and high flows appear to increase Back Close in these model runs for these gauges while allowing these changes to be quantified. Full Screen / Esc

25 3.3.4 Decadal percentile analysis Printer-friendly Version

The annual timeseries shown in Fig.5 is very variable and systematic changes Interactive Discussion throughout the century could be masked by this variability therefore in this section the 10th and 90th percentiles for each decade and each model run are considered to 5808 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion see if there is any systematic change on a decadal basis through to 2100. There is little difference between the two models for the 10th percentile (not shown) for most of HESSD the gauges, this is mainly due to the very low river flows at the lowest flow times of the 12, 5789–5840, 2015 year. Only the Pandu and Bahadurabad gauges on the Brahmaputra and the Farakka 5 gauge on the Ganges show a non-zero value for the lowest 10 % of river flows through to the 2100s. These three gauges indicate a slight increase for the 10th percentile for South Asia river flow each decade through to 2100. projections and their Figure 11 shows the 90th percentile for both models calculated for each decade implications for water from 1970 to 2100 for each of the river gauges specified in Table1. The 90th resources 10 percentile values (Fig. 11) are generally much more variable than those for the 10th percentile, particularly in terms of changes through to the 2100s. Considering the C. Mathison et al. gauges according to their location across the Himalayan arc from west to east, the HadCM3 simulation projects an increase in the flow for the two gauges on the Indus Title Page (Attock and Kotri gauges, shown in Fig. 11, 1st row) and the Chenab-Panjnad gauge 15 (Fig. 11, 2nd row, left column), however ECHAM5 is generally indicating a much flatter Abstract Introduction trajectory or decreasing river flow on these rivers. Conclusions References The gauges located toward the middle of the Himalayan arc in this analysis; namely Bhagirathi-Tehri (Fig. 11, 2nd row, right column), Karnali river gauges – Benighat Tables Figures and Chisapani (Fig. 11, 3rd row), Narayani-Devghat and Arun-Turkeghat (Fig. 11, 4th 20 row) generally show increases across the decades to 2100 in both models. There is J I very close agreement between the two simulations for the Narayani-Devghat, Arun- Turkeghat (Fig. 11, 4th row) and Bhagirathi-Tehri (Fig. 11, 2nd row, right column) J I gauges with the former two showing less variability between decades than the others in Back Close the analysis. The Karnali-Benighat gauge (Fig. 11, 3rd row, left column) also has less Full Screen / Esc 25 variability between the decades, however there is a systematic difference between the two simulations that remains fairly constant across the decades. The Karnali-Chisapani Printer-friendly Version gauge (Fig. 11, 3rd row, right column) has the largest variability between simulations and decades of the models in the analysis that are most central on the Himalayan arc, Interactive Discussion

5809 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion this gauge still shows an increase overall in both models although the gradient of this increase is smaller for ECHAM5 than HadCM3. HESSD The Farakka-Ganges gauge (Fig. 11, 5th row, left column) and the Brahmaputra 12, 5789–5840, 2015 gauges – Bahadurabad and Pandu (Fig. 11,5th row, right column and 6th row, right 5 column, respectively), represent the most easterly river gauges in the analysis; these gauges show an increase in both simulations through to the 2100s, although this is South Asia river flow more pronounced in ECHAM5 than HadCM3 for the Brahmaputra gauges. There is projections and their much closer agreement between the two simulations at the Farakka-Ganges gauge implications for water (Fig. 11, 5th row, left column) which is located slightly further west than the two resources 10 Brahmaputra gauges. This analysis suggests that neither simulation is consistently showing a systematic C. Mathison et al. increase in the 90th percentile of river flows across all the gauges, however it does highlight the different behaviour in the two simulations across the Himalayas. The Title Page HadCM3 simulation shows increases in western river flows which are not evident in 15 the ECHAM5 simulation; this may be explained by the HadCM3 simulation depicting Abstract Introduction an increase in the occurance of western disturbances and an increase in total snowfall Conclusions References which is not evident in the ECHAM5 simulation (Ridley et al., 2013). In contrast, for the eastern gauges, both simulations show an increase in river flow, although the ECHAM5 Tables Figures simulation shows larger increases than HadCM3. The central gauges suggested 20 a more mixed result, with the models more in agreement with each other; this may be J I due to the reducing influence of the western disturbances in the HadCM3 simulation from west to east across the Himalaya therefore resulting in smaller differences J I between the the two simulations at these gauges. Back Close

Full Screen / Esc 4 Implications of changes in future river ﬂows Printer-friendly Version

25 In this section we consider the implications of the projected future changes in river flows Interactive Discussion for South Asia on water resources, the key points from this discussion are summarised in Table3. In the present day water resources in South Asia are complicated, 5810 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion precariously balanced between receiving some of the largest volumes of precipitation in the world and therefore the frequent risk of flooding and yet regularly enduring HESSD water shortages. The complexity is increased by the competition between states and 12, 5789–5840, 2015 countries for resources from rivers that flow large distances crossing state and country 5 borders each with their own demands on resource. Annually India receives about 4000 km3 of precipitation with 3000 km3 falling during the ASM period. A proportion, South Asia river flow estimated to be just over 45 % of this precipitation (approximately 1869 km3), finds projections and their its way into the river and replenishable groundwater system (Gupta and Deshpande, implications for water 2004) which form the basis for the water resources of the country. Of the water that resources 10 actually finds its way into the system only 60 % of it is currently put to beneficial use, in terms of volume; this is approximately 690 km3 of surface water and 433 km3 of ground C. Mathison et al. water (Aggarwal et al., 2012). This means there is a gap between the amount of water resource flowing through South Asia and the actual useable amount, for example the Title Page total flow for the Brahmaputra basin is approximately 629 km3 of which only 24 km3 is 15 usable (Kumar et al., 2005). There is therefore huge potential for improvements in the Abstract Introduction efficiency of systems for irrigation and the domestic water supply that could ease some Conclusions References of the pressures on water resources currently experienced already and predicted in the future for some areas as the demand for water increases. Tables Figures In the last 50 years there have already been efficiency improvements, such as 20 development of irrigation systems and use of high yielding crop varieties that have J I fuelled the rapid development in agriculture across South Asia making the region more self-sustained and alleviating poverty (Kumar et al., 2005); however this has had J I a large impact on the regions river ecosystems resulting in habitat loss and reduced Back Close biodiversity (Sarkar et al., 2012). Vörösmarty et al.(2010) find that in developing Full Screen / Esc 25 regions, where investment in water infrastructure is low and water security is threatened there tends to be a coincident risk of biodiversity loss, with the main threat due to water Printer-friendly Version resource development and increased pollution from the use of pesticides and fertilizer. 3 Gupta and Deshpande(2004) estimate that a minimum storage of 385 km is needed Interactive Discussion across all the basins in India to balance seasonal flows and irrigate 760 000 km2

5811 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion although how this translates to an individual river in terms of the river flows needed to maintain ecosystems and biodiversity (also referred to as environmental flows) is HESSD a complex problem. Historically arbitrary thresholds based on a percentage of the 12, 5789–5840, 2015 annual mean flow have been used to estimate minimum flows, but these simplistic 5 estimates do not take account of the flow variability that is crucial for sustaining river ecosystems (Arthington et al., 2006; Smakhtin et al., 2006). Environmental flows are South Asia river flow defined by Smakhtin and Anputhas(2006) as the ecologically acceptable flow regime projections and their designed to maintain a river in an agreed or predetermined state. The variability in implications for water river flows through the year have important ecological significance; for example low resources 10 flows are important for algae control and therefore maintaining water quality, while high flows are important for wetland flooding and preserving the river channel. When C. Mathison et al. considering the implications of future changes in climate on river flows and therefore surface water resources, an estimate of the environmental requirement, both in terms Title Page of the flow variability as well as the minimum flows, are an important consideration. 15 These important ecological thresholds together with the flows which cause inundation Abstract Introduction and crop damage have been calculated for individual basins, such as the lower Conclusions References Brahmaputra river basin by Gain et al.(2013) and the East Rapti River in Nepal by Smakhtin et al.(2006), however they are not easily quantified in general terms for Tables Figures different rivers with many methods requiring calibration for applications to different 20 regions. J I In India the domestic requirement for water is the highest priority but is only 5 % of the total demand (this equates to approximately 30 km3 of which 17 km3 is from surface J I water and the rest groundwater). Irrigation is the second highest priority accounting for Back Close a much greater proportion, approximately 80 % of India’s total demand for water; this 3 3 3 Full Screen / Esc 25 equates to more than 520 km with 320 km from surface water and 206 km from groundwater (Kumar et al., 2005). Biemans et al.(2013) study future water resources Printer-friendly Version for food production using LPJml and the HNRCMS. The LPJml simulated extraction 3 −1 varies considerably between basins; the largest occuring in the Indus (343 km year ) Interactive Discussion followed by the Ganges (281 km3 year−1) and Brahmaputra (45 km3 year−1). The

5812 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion Brahmaputra has the smallest percentage of irrigated crop production (approximately 40 %) followed by the Ganges (less than 75 %) and the Indus where more than 90 % HESSD of crop production is on irrigated land. The Indus has the largest proportion of water 12, 5789–5840, 2015 sourced from rivers and lakes of the three basins. LPJml also simulates ground water 5 extractions Biemans et al.(2013) these are thought to be important for the Indus and parts of the Ganges but not the Brahmaputra. The model simulations presented in South Asia river flow this analysis do not explicitly include groundwater, primarily focusing on river flows projections and their and therefore the surface water component of resource for this region. There is also implications for water no irrigation included in these simulations, which could be important particularly on resources 10 the basin scale. The impacts of extensive irrigation on the atmosphere are complex but could have a positive impact on water availability (Harding et al., 2013) due to C. Mathison et al. evaporation and water being recycled within the basin, for example, Tuinenburg et al. (2014) estimate that up to 35 % of additional evaporation is recycled within the Ganges Title Page basin. 15 In general the analysis here shows that the magnitudes of the higher river flows could Abstract Introduction increase for these gauges (see Table1), in some cases these increases are above the Conclusions References range of variability used for this analysis (1.5 SD). While this could be positive in terms of surface water resources for irrigation, the potential changes seem to occur during the Tables Figures ASM season and therefore when river flow is at its maximum; therefore this increase 20 may not be critical for water resources but could still be beneficial where there is the J I capacity to store the additional flow for use during periods of low flow. Additional water storage capacity for example through rainwater harvesting, could greatly increase the J I useable water resource for the Ganges–Brahmaputra catchments (Kumar et al., 2005) Back Close and potentially alleviate the increased risk of flooding during the ASM when rainfall is Full Screen / Esc 25 most persistent and rivers are already at their peak flow. South Asia, even in the current climate, is particularly susceptible to flooding due to the high temporal and spatial Printer-friendly Version variability of rainfall of the region, for example approximately 20 % of Bangladesh floods annually (Mirza, 2002). Several studies have highlighted increases in both the extremes Interactive Discussion (Sharma, 2012; Rajeevan et al., 2008; Goswami et al., 2006; Joshi and Rajeevan,

5813 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion 2006) and the variability (Gupta et al., 2005) of precipitation in recent years, where extreme rainfall events have resulted in catastrophic levels of river flooding. Over HESSD 30 million people in India alone are affected by floods and more than 1500 lives are 12, 5789–5840, 2015 lost each year (Gupta et al., 2003), the economic cost of flooding is also considerable 5 with the cumulative flood related losses estimated to be of the order of USD 16 billion between 1978 and 2006 (Singh and Kumar, 2013). South Asia river flow The timing of the peak flows of major rivers in this region is also very important in projections and their terms of flooding. In 1998 the peak flows of the Ganges and the Brahmaputra rivers implications for water occurred within 2 days of each other resulting in devastating flooding across the entire resources 10 central region of Bangladesh inundating aproximately 70 % of the country, the flood waters then remained above danger levels for more than 60 days (Mirza, 2002). This C. Mathison et al. event caused extensive loss of life and livelihood in terms of damaged crops, fisheries and property with the slow recedance of flood waters hindering the relief operation and Title Page recovery of the region. This analysis does not suggest any change to the timing of the 15 peak flows, only the magnitude, however given the high probability of two rivers in this Abstract Introduction region having coincident peak flows in any given year (Mirza, 2002) and the likelihood Conclusions References that severe flooding will result, means that an increase in the magnitude of the peak could still be significant. Flooding can have a large impact on crops, for example in Tables Figures Bangladesh over 30 % of the total flood related damages are due to the loss of crops; 20 the estimated crop damage from the 1998 floods was estimated to be 3.0 million t (Gain J I et al., 2013). Slow receeding of flood water can also mean the ground is not in a suitable condition to sow the next crop, restricting the growing time and potentially affecting crop J I yields for the following year. Back Close Another proposed though controversial method aimed at alleviating flooding in the Full Screen / Esc 25 South Asia region is inter-basin transfer through the National River Linking Project (NRLP); this is an attempt to redistribute the water between rivers by linking those Printer-friendly Version rivers with a surplus to those with a deficit (Gupta and Deshpande, 2004). The success of these projects depends on the elevation of the catchment providing the water being Interactive Discussion above that of the receiving catchment so catchments with a low elevation such as

5814 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion the Brahmaputra can only transfer a small amount despite having large problems with flooding. On the other hand a limited amount of flooding could also be a benefit, HESSD particularly for rice crops, as the inundation of clear water benefits crop yield due to 12, 5789–5840, 2015 the fertilization effect of nitrogen producing blue-green algae in the water (Mirza et al., 5 2003). In these simulations the occurrance of the lowest flows potentially reduces in the South Asia river flow future, which could translate into an increase in the surface water resource in this projections and their region, for periods when the river flows are traditionally very low and water is usually implications for water scarce. This could mean that the current and increasing pressure on ground water resources 10 (Rodell et al., 2009) may be alleviated in future years. Alternatively increases in the lowest flows may enable adaptation to a changing climate and the modification of C. Mathison et al. irrigation practises. Current projections of future climate suggest that temperatures could also increase for this region (Cruz et al., 2007), this poses a threat to crop Title Page yields of a different kind because this is a region where temperatures are already at 15 a physiological maxima for some crops (Gornall et al., 2010). Rice yield, for example, Abstract Introduction is adversely affected by temperatures above 35 ◦C at the critical flowering stage of Conclusions References its development (Yoshida, 1981) and wheat yields could be also affected by rising ◦ −1 temperatures, with estimated losses of 4–5 milliont( C) temperature rise through Tables Figures the growing period (Aggarwal et al., 2012). Additional water resource for irrigation at 20 previously low flow times of the year could allow sowing to take place at a different time J I of the year in order to avoid the highest temperatures, thereby reducing the likelihood of crop failure. However with increasing variability and extremes, a potential feature J I of the future climate for this region, there is also the increased risk of longer periods Back Close with below average rainfall and potentially more incidences of drought; this could lead Full Screen / Esc 25 to additional demand for water for irrigation to prevent crops becoming water stressed (Aggarwal et al., 2012). There may also be increases in demand from other sources Printer-friendly Version other than agriculture, for example the increasing population (United Nations, 2013) or the reduced availability of ground water of an acceptable quality for domestic use Interactive Discussion (Gregory et al., 2005). Any of these factors, either individually or combined, could

5815 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion effectively cancel out any or all increases in resource from increased river flow due to climate change. HESSD 12, 5789–5840, 2015 5 Conclusions South Asia river flow In this analysis the first 25 km resolution regional climate projections of riverflow for projections and their 5 the South Asia region are presented. A sub-selection of the HNRCMs are used to implications for water provide runoff to a river routing model in order to provide river flow rate which can be resources compared directly with a downscaled ERAint simulation and any available observation data for river basins in the South Asia region. This analysis focuses on the major South C. Mathison et al. Asia river basins which originate in the glaciated Hindu-Kush Karakoram Himalaya; 10 Ganges/Brahmaputra and the Indus. The aim of this analysis is two-fold; firstly to understand the river flows in the RCM in the two simulations and how useful they Title Page are for understanding the changes in water resources for South Asia and secondly to understand what the projected changes in river flow to the 2100s might mean for water Abstract Introduction resources across the Himalaya region. Conclusions References 15 The two simulations in this analysis cannot capture the full range of variability, however the two GCMs that are downscaled using this RCM do capture a range Tables Figures of temperatures and variability in precipitation similar to the AR4 ensemble for Asia (Christensen et al., 2007) which is for a much larger domain than the HighNoon J I domain analysed here (Mathison et al., 2013). A number of GRDC gauge stations J I 20 (GRDC, 2014), selected to capture the range of conditions across the Himalayan Back Close arc and sample the major river basins, provide the observations of river flow for comparison against the simulations. The lack of recent river flow data limited the Full Screen / Esc gauges that could be selected for analysis, however using the downscaled ERAint simulation provides a constrained estimate of the South Asia water cycle in the Printer-friendly Version 25 absence of robust observations and is used in addition to the observations to provide Interactive Discussion a useful benchmark against which to compare the downscaled GCM simulations. In general there is a tendancy for overestimation of river flow rate across the selected 5816 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion gauges compared to the GRDC observations, however comparison against the ERAint simulation is more mixed with some gauges showing higher and others lower river HESSD flows for the downscaled GCMs compared with ERAint. However in general most of the 12, 5789–5840, 2015 simulations broadly agree with observations and ERAint to within the range of natural 5 variability (chosen to be 1.5 SD for this analysis) and agree on the periods of highest and lowest river flow, indicating that the RCM is able to capture the main features of South Asia river flow both the climate and hydrology of the region. projections and their The simulations suggest that the annual average river flow is increasing toward the implications for water 2100s, although these trends are often masked by the large inter-annual variability resources 10 of river flows in this region, for some of the gauges the river flow rates are almost doubled by the end of the century. These increases in river flows are reflected in the C. Mathison et al. seasonal cycle for the two future periods (2050s and 2080s) which indicate that most of the changes occur during peak flow periods with some gauges showing changes Title Page above the range of present day natural variability. These gauges tend to be toward 15 the middle of the Himalayan arc, so this could be due to the increasing influence Abstract Introduction of the ASM and reducing influence of western disturbances from west to east. The Conclusions References gauges located furthest west and east in this analysis seem to lie within the present day natural variability. The analysis shown here does not suggest a systematic change Tables Figures in the models for the timing of the maximum and minimum river flows relative to 20 the present day suggesting an over all increase in water resources at the top and J I bottom of the distribution. This has positive and negative implications with potentially more resource during usually water scarce periods but also carries implications for an J I already vulnerable population in terms of increased future flood risk during periods Back Close where the river flow is particularly high. Bangladesh is particularly susceptible to Full Screen / Esc 25 flooding, therefore any increase in maximum flows for rivers in this region could be important in terms of loss of life, livelihoods, particularly agriculture and damage to Printer-friendly Version infrastructure. Historically management policies for rivers in this region have focussed on percentage of the average annual flow which does not take into account the Interactive Discussion

5817 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion importance of flow variability as well as minimum flows, which are important for sustaining river ecosystems. HESSD While this analysis suggests a general increase in potential water resources from 12, 5789–5840, 2015 rivers for this region to 2100 due to climate change, there are a number of factors which 5 could have a larger effect on water resources for this region and effectively cancel out any increase. For example rising population, depletion of ground water, increases in South Asia river flow demand for water from sources other than agriculture. In addition increasing variability projections and their of the South Asia climate could lead to long periods with below average rainfall which implications for water could also increase the demand for irrigation. Further more the results shown here do resources 10 not currently explicitly include the glacial contribution to river flow for these catchments and gauges. Including glacial processes in the form of a glacier model together with C. Mathison et al. river routing within the land-surface representation will be useful to establish if the contribution from glaciers changes the timing and/or magnitude of both the lowest and Title Page highest flows in these gauges. Likewise including representation of water extraction 15 (both from rivers and groundwater) particularly for irrigation, the biggest user of water Abstract Introduction in the region, will help to provide a more complete picture of the water resources for Conclusions References the South Asia region. Understanding the interactions between availability of water resources, irrigation and food production for this region by using a more integrated Tables Figures approach, such as that used in Biemans et al.(2013) may also help with understanding 20 how pressures on resources could change with time. In support of this work and others, J I there is also a need for good quality observations of both precipitation and river flow that is available for long enough time periods to conduct robust water resource assessments J I for this region. Back Close

Acknowledgements. The research leading to these results has received funding from the Full Screen / Esc 25 European Union Seventh Framework Programme FP7/2007-2013 under grant agreement no 603864. Camilla Mathison, Pete Falloon and Andy Wiltshire were supported by the Joint UK Printer-friendly Version DECC/Defra Met Oﬃce Hadley Centre Climate Programme (GA01101). Thanks to Neil Kaye for his GIS expertise. Interactive Discussion

5818 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion References HESSD Aggarwal, P., Talukdar, K., and Mall, R.: Potential Yields of Rice-Wheat System in the Indo- Gangetic Plains of India, Rice-Wheat Consortium for the Indo-Gangetic Plains, New Delhi, 12, 5789–5840, 2015 India, Rice-Wheat Consortium Paper Series, 10, 16, 2000. 5793 5 Aggarwal, P.,Palanisami, K., Khanna, M., and Kakumanu, K.: Climate change and food security of India: adaptation strategies for the irrigation sector, World Agr., 3, 20–26, 2012. 5792, South Asia river flow 5811, 5815 projections and their Akhtar, M., Ahmad, N., and Booij, M.: The impact of climate change on the water resources implications for water of Hindukush–Karakorum–Himalaya region under different glacier coverage scenarios, J. resources 10 Hydrol., 355, 148–163, doi:10.1016/j.jhydrol.2008.03.015, 2008. 5793 Andermann, C., Bonnet, S., and Gloaguen, R.: Evaluation of precipitation data sets along C. Mathison et al. the Himalayan front, Geochem. Geophy. Geosy., 12, Q07023, doi:10.1029/2011GC003513, 2011. 5795 Annamalai, H., Hamilton, K., and Sperber, K.: The South Asian summer monsoon and Title Page 15 its relationship with ENSO in the IPCC AR4 simulations, J. Climate, 20, 1071–1092, doi:10.1175/JCLI4035.1, 2007. 5793, 5797 Abstract Introduction Arthington, A. H., Bunn, S. E., Poff, N. L., and Naiman, R. J.: The challenge of providing environmental flow rules to sustain river ecosystems, Ecol. Appl., 16, 1311–1318, Conclusions References doi:10.1890/1051-0761(2006)016[1311:TCOPEF]2.0.CO;2, 2006. 5812 Tables Figures 20 Bajracharya, S. and Shrestha, B.: The Status of Glaciers in the Hindu Kush-Himalayan Region, Kathmandu: ICIMOD, available at: http://lib.icimod.org/record/9419, last access: November 2014, 2011. 5796 J I Barnett, T., Adam, J., and Lettenmaier, D.: Potential impacts of a warming climate on water J I availability in snow-dominated regions, Nature, 438, 303–309, doi:10.1038/nature04141, 25 2005. 5792 Back Close Biemans, H., Speelman, L., Ludwig, F., Moors, E., Wiltshire, A., Kumar, P., Gerten, D., and Kabat, P.: Future water resources for food production in five South Asian river basins and Full Screen / Esc potential for adaptation – a modeling study, Sci. Total Environ., 468–469, S117–S131, doi:10.1016/j.scitotenv.2013.05.092, 2013. 5799, 5803, 5812, 5813, 5818 Printer-friendly Version 30 Christensen, J., Hewitson, B., Busuioc, A., Chen, A., Gao, X., Held, I., Jones, R., Kolli, R., Kwon, W.-T., Laprise, R., na Rueda, V. M., Mearns, L., Meneńdez, C., Räisänen, J., Rinke, A., Interactive Discussion Sarr, A., and Whetton, P.: Regional climate projections, in: Climate Change 2007: the 5819 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion Physical Science Basis. Contribution of Working Group I, Fourth Assessment Report of the Intergovernmental Panel on Climate Change, available at: https://www.ipcc.ch/publications_ HESSD and_data/ar4/wg1/en/ch11.html, last access: October 2014, 2007. 5792, 5797, 5800, 5816 Clark, D. B. and Gedney, N.: Representing the effects of subgrid variability of soil 12, 5789–5840, 2015 5 moisture on runoff generation in a land surface model, J. Geophys. Res., 113, D10111, doi:10.1029/2007JD008940, 2008. 5798 Collins, D. N., Davenport, J. L., and Stoffel, M.: Climatic variation and runoff from partially- South Asia river flow glacierised Himalayan tributary basins of the Ganges, Sci. Total Environ., 468–469, S48– projections and their S59, doi:10.1016/j.scitotenv.2013.10.126, 2013. 5794 implications for water 10 Cruz, R., Harasawa, H., Lal, M., Wu, S., Anokhin, Y., Punsalmaa, B., Honda, Y., Jafari, M., resources Li, C., and Huu Ninh, N.: Asia. Climate change 2007: impacts, adaptation and vulnerability, in: Contribution of Working Group II, Fourth Assessment Report of the Intergovernmental Panel C. Mathison et al. on Climate Change, 469–506, available at: https://www.ipcc.ch/publications_and_data/ar4/ wg2/en/ch10.html, last access: October 2014, 2007. 5815 15 Dhar, O. and Nandargi, S.: A study of floods in the Brahmaputra basin in India, Int. J. Climatol., Title Page 20, 771–781, doi:10.1002/1097-0088(20000615)20:7<771::AID-JOC518>3.0.CO;2-Z, 2000. 5796 Abstract Introduction Dimri, A., Yasunari, T., Wiltshire, A., Kumar, P., Mathison, C., Ridley, J., and Jacob, D.: Application of regional climate models to the Indian winter monsoon over the western Conclusions References 20 Himalayas, Sci. Total Environ., 468, S36–S47, 2013. 5794 Tables Figures Dolman, A. J. and Gregory, D.: The parametrization of rainfall interception in GCMs, Q. J. Roy. Meteor. Soc., 118, 455–467, doi:10.1002/qj.49711850504, 1992. 5798 Essery, R. L. H., Best, M. J., and Cox, P. M.: MOSES 2.2 technical documentation, J I Hadley Centre Technical Note, 30, available at: http://www.metoffice.gov.uk/archive/ J I 25 hadley-centre-technical-note-30, last access: February 2015, 2001. 5798 Essery, R. L. H., Best, M. J., Betts, R. A., Cox, P. M., and Taylor, C.: Explicit representation Back Close of subgrid heterogeneity in a GCM land surface scheme, J. Hydrometeorol., 4, 530–543, doi:10.1175/1525-7541(2003)004<0530:EROSHI>2.0.CO;2, 2003. 5797, 5798 Full Screen / Esc Falloon, P., Betts, R., and Bunton, C.: New global river routing scheme in the unified 30 model, Hadley Centre Technical Note, 72, available at: http://www.metoffice.gov.uk/archive/ Printer-friendly Version hadley-centre-technical-note-72, last access: February 2015, 2007. 5799 Interactive Discussion

5820 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion Falloon, P., Betts, R. A., Wiltshire, A., Dankers, R., Mathison, C., McNeall, D., Bates, P., and Trigg, M.: Validation of river flows in hadgem1 and hadcm3 with the trip river flow model, J. HESSD Hydrometeorol., 12, 1157–1180, doi:10.1175/2011JHM1388.1, 2011. 5793, 5803 Fekete, B. M., Vörösmarty, C. J., and Lammers, R. B.: Scaling gridded river networks for 12, 5789–5840, 2015 5 macroscale hydrology: development, analysis, and control of error, Water Resour. Res., 37, 1955–1967, doi:10.1029/2001WR900024, 2001. 5798 Fujita, K. and Nuimura, T.: Spatially heterogeneous wastage of Himalayan glaciers, P. Natl. South Asia river flow Acad. Sci. USA, 108, 14011–14014, 2011. 5792 projections and their Gain, A. K., Apel, H., Renaud, F. G., and Giupponi, C.: Thresholds of hydrologic flow regime of implications for water 10 a river and investigation of climate change impact the case of the Lower Brahmaputra river resources Basin, Climatic Change, 120, 463–475, doi:10.1007/s10584-013-0800-x, 2013. 5812, 5814 Gedney, N. and Cox, P.:The sensitivity of global climate model simulations to the representation C. Mathison et al. of soil moisture heterogeneity, J. Hydrometeorol., 4, 1265–1275, doi:10.1175/1525- 7541(2003)004<1265:TSOGCM>2.0.CO;2, 2003. 5798 15 Gordon, C., Cooper, C., Senior, C. A., Banks, H., Gregory, J. M., Johns, T. C., Mitchell, J. F. B., Title Page and Wood, R. A.: The simulation of SST, sea ice extents and ocean heat transports in a version of the Hadley Centre coupled model without flux adjustments, Clim. Dynam., 16, Abstract Introduction 147–168, doi:10.1007/s003820050010, 2000. 5797 Gornall, J., Betts, R., Burke, E., Clark, R., Camp, J., Willett, K., and Wiltshire, A.: Implications Conclusions References 20 of climate change for agricultural productivity in the early twenty-first century, Philos. T. Roy. Tables Figures Soc. B, 365, 2973–2989, doi:10.1098/rstb.2010.0158, 2010. 5815 Goswami, B. and Xavier, P. K.: Dynamics of “internal” interannual variability of the Indian summer monsoon in a GCM, J. Geophys. Res., 110, D24104, doi:10.1029/2005JD006042, J I 2005. 5792, 5800 J I 25 Goswami, B. N., Venugopal, V., Sengupta, D., Madhusoodanan, M. S., and Xavier, P. K.: Increasing trend of extreme rain events over India in a warming environment, Science, 314, Back Close 1442–1445, doi:10.1126/science.1132027, 2006. 5813 GRDC: The Global Runoff Data Centre, 56002 Koblenz, Germany, 2014. 5794, 5795, 5802, Full Screen / Esc 5816 30 Gregory, P., Ingram, J., and Brklacich, M.: Climate change and food security, Philos. T. R. Soc. Printer-friendly Version B, 360, 2139–2148, doi:10.1098/rstb.2005.1745, 2005. 5792, 5815 Interactive Discussion

5821 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion Gupta, A. D., Babel, M. S., Albert, X., and Mark, O.: Water sector of Bangladesh in the context of integrated water resources management: a review, Int. J. Water Resour. D., 21, 385–398, HESSD doi:10.1080/07900620500037818, 2005. 5814 Gupta, S. and Deshpande, D.: Water for India in 2050: first order assessment of available 12, 5789–5840, 2015 5 options, Curr. Sci. India, 86, 1216–1224, 2004. 5792, 5811, 5814 Gupta, S., Javed, A., and Datt, D.: Economics of flood protection in India, Nat. Hazards, 28, 199–210, 2003. 5814, 5829 South Asia river flow Harding, R., Blyth, E., Tuinenburg, O., and Wiltshire, A.: Land atmosphere feedbacks and their projections and their role in the water resources of the Ganges basin, Sci. Total Environ., 468–469, S85–S92, implications for water 10 doi:10.1016/j.scitotenv.2013.03.016, 2013. 5796, 5799, 5813 resources Immerzeel, W., van Beek, L., and Bierkens, M.: Climate change will affect the Asian water towers, Science, 328, 1382–1385, doi:10.1126/science.1183188, 2010. 5792 C. Mathison et al. Johnston, R. and Smakhtin, V.: Hydrological Modeling of Large river Basins: How Much is Enough?, Water Resour. Manag., 28, 2695–2730, doi:10.1007/s11269-014-0637-8, 2014. 15 5795 Title Page Jones, R. G., Noguer, M., Hassell, D. C., Hudson, D., Wilson, S. S., Jenkins, G. J., and Mitchell, J. F.: Generating high resolution climate change scenarios using PRECIS, Met Abstract Introduction Office Hadley Centre, Exeter, UK, 0–40, available at: http://www.metoffice.gov.uk/media/pdf/ 6/5/PRECIS_Handbook.pdf, last access: February 2015, 2004. 5797 Conclusions References 20 Joshi, U. and Rajeevan, M.: Trends in precipitation extremes over India, National Climate Tables Figures Centre Research report, 3, available at: http://www.imdpune.gov.in/ncc_rept/RESEARCH% 20%20REPORT%203.pdf, last access: February 2015, 2006. 5813 Kapnick, S. B., Delworth, T. L., Ashfaq, M., Malyshev, S., and Milly, P. C. D.: Snowfall less J I sensitive to warming in Karakoram than in Himalayas due to a unique seasonal cycle, Nat. J I 25 Geosci., 7, 834–840, doi:10.1038/ngeo2269, 2014. 5794 Kumar, P., Wiltshire, A., Mathison, C., Asharaf, S., Ahrens, B., Lucas-Picher, P., Back Close Christensen, J. H., Gobiet, A., Saeed, F., Hagemann, S., and Jacob, D.: Down- scaled climate change projections with uncertainty assessment over India using Full Screen / Esc a high resolution multi-model approach, Sci. Total Environ., 468–469, S18–S30, 30 doi:10.1016/j.scitotenv.2013.01.051, 2013. 5797, 5799, 5800 Printer-friendly Version Kumar, R., Singh, R., and Sharma, K.: Water resources of India, Curr. Sci. India, 89, 794–811, 2005. 5792, 5811, 5812, 5813 Interactive Discussion

5822 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion Linsley, R., Kohler, M., and Paulhus, J.: Hydrology for Engineers, McGraw-Hill series in water resources and environmental engineering, McGraw-Hill, available at: http://books.google.co. HESSD uk/books?id=9vROAAAAMAAJ, last access: February 2015, 1982. 5798 Lucas-Picher, P., Christensen, J. H., Saeed, F., Kumar, P., Asharaf, S., Ahrens, B., 12, 5789–5840, 2015 5 Wiltshire, A. J., Jacob, D., and Hagemann, S.: Can regional climate models represent the Indian Monsoon?, J. Hydrometeorol., 12, 849–868, doi:10.1175/2011JHM1327.1, 2011. 5793, 5799 South Asia river flow Mathison, C., Wiltshire, A., Dimri, A., Falloon, P., Jacob, D., Kumar, P., Moors, E., projections and their Ridley, J., Siderius, C., Stoffel, M., and Yasunari, T.: Regional projections of implications for water 10 North Indian climate for adaptation studies, Sci. Total Environ., 468–469, S4–S17, resources doi:10.1016/j.scitotenv.2012.04.066, 2013. 5792, 5793, 5797, 5799, 5800, 5801, 5806, 5816 McMillan, H., Freer, J., Pappenberger, F., Krueger, T., and Clark, M.: Impacts of uncertain river C. Mathison et al. flow data on rainfall–runoff model calibration and discharge predictions, Hydrol. Process., 24, 1270–1284, 2010. 5803 15 Mirza, M. M. Q.: Global warming and changes in the probability of occurrence of Title Page floods in Bangladesh and implications, Global Environmental Change, 12, 127–138, doi:10.1016/S0959-3780(02)00002-X, 2002. 5813, 5814 Abstract Introduction Mirza, M. M. Q., Warrick, R., Ericksen, N., and Kenny, G.: Trends and persistence in precipitation in the Ganges, Brahmaputra and Meghna river basins, Hydrological Sciences Conclusions References 20 Journal, 43, 845–858, doi:10.1080/02626669809492182, 1998. 5795 Tables Figures Mirza, M. M. Q., Warrick, R. A., and Ericksen, N. J.: The Implications of Climate Change on Floods of the Ganges, Brahmaputra and Meghna Rivers in Bangladesh, Climatic Change, 57, 287–318, doi:10.1023/A:1022825915791, 2003. 5815 J I Moors, E. J., Groot, A., Biemans, H., van Scheltinga, C. T., Siderius, C., Stoffel, M., Huggel, C., J I 25 Wiltshire, A., Mathison, C., Ridley, J., Jacob, D., Kumar, P., Bhadwal, S., Gosain, A., and Collins, D. N.: Adaptation to changing water resources in the Ganges basin, northern India, Back Close Environ. Sci. Policy, 14, 758–769, doi:10.1016/j.envsci.2011.03.005, 2011. 5797 Morse, A., Prentice, C., and Carter, T.: Assessments of climate change impacts, Ensembles: Full Screen / Esc Climate change and its impacts: Summary of research and results from the ENSEMBLES 30 project, 107–129, available at: http://ensembles-eu.metoffice.com/docs/Ensembles_final_ Printer-friendly Version report_Nov09.pdf, last access: February 2015, 2009. 5799 Interactive Discussion

5823 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion Nohara, D., Kitoh, A., Hosaka, M., and Oki, T.: Impact of climate change on river discharge projected by multimodel ensemble, J. Hydrometeorol., 7, 1076–1089, HESSD doi:10.1175/JHM531.1, 2006. 5800 Oki, T. and Sud, Y. C.: Design of Total Runoff Integrating Pathways (TRIP). 12, 5789–5840, 2015 5 A global river channel network, Earth Interact., 2, 1–37, doi:10.1175/1087- 3562(1998)002<0001:DOTRIP>2.3.CO;2, 1998. 5793, 5798 Oki, T., Nishimura, T., and Dirmeyer, P.: Assessment of annual runoff from land surface South Asia river flow models using Total Runoff Integrating Pathways (TRIP), J. Meteorol. Soc. Jpn., 77, 235–255, projections and their doi:10.1080/02626668509490989, 1999. 5799, 5803 implications for water 10 Pope, V., Gallani, M. L., Rowntree, P. R., and Stratton, R. A.: The impact of new physical resources parametrizations in the Hadley Centre climate model: HadAM3, Clim. Dynam., 16, 123–146, doi:10.1007/s003820050009, 2000. 5797 C. Mathison et al. Rajeevan, M., Bhate, J., and Jaswal, A. K.: Analysis of variability and trends of extreme rainfall events over India using 104 years of gridded daily rainfall data, Geophys. Res. Lett., 35, 15 L18707, doi:10.1029/2008GL035143, 2008. 5813 Title Page Ridley, J., Wiltshire, A., and Mathison, C.: More frequent occurrence of westerly disturbances in Karakoram up to 2100, Sci. Total Environ., 468–469, S31–S35, Abstract Introduction doi:10.1016/j.scitotenv.2013.03.074, 2013. 5794, 5810 Rodell, M., Velicogna, I., and Famiglietti, J.: Satellite-based estimates of groundwater depletion Conclusions References 20 in India, Nature, 460, 999–1002, doi:10.1038/nature08238, 2009. 5792, 5815 Tables Figures Roeckner, E., Bäuml, G., Bonaventura, L., Brokopf, R., Esch, M., Giorgetta, M., Hagemann, S., Kirchner, I., Kornblueh, L., Manzini, E., Rhodin, A., Schlese, U., Schulzweida, U., and Tompkins, A.: The atmospheric general circulation model ECHAM 5. PART I: Model J I description, Max Planck Institute for Meteorology Rep. 349, available at: http://www.mpimet. J I 25 mpg.de/fileadmin/publikationen/Reports/max_scirep_349.pdf, last access: February 2015, 2003. 5797 Back Close Sarkar, U. K., Pathak, A. K., Sinha, R. K., Sivakumar, K., Pandian, A. K., Pandey, A., Dubey, V. K., and Lakra, W. S.: Freshwater fish biodiversity in the River Ganga (India): Full Screen / Esc changing pattern, threats and conservation perspectives, Rev. Fish Biol. Fisher., 22, 251– 30 272, doi:10.1007/s11160-011-9218-6, 2012. 5811 Printer-friendly Version Scott, D.: Multivariate Density Estimation: Theory, Practice, and Visualization, vol. 383, mainly chapter 6, John Wiley & Sons, USA, 2009. 5806 Interactive Discussion

5824 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion Sharma, D.: Situation analysis of flood disaster in south and southeast Asia – a need of integrated approach, International Journal of Science, Environment and Technology, 1, 167– HESSD 173, 2012. 5813 Silverman, B. W.: Density estimation for statistics and data analysis, Monographs on Statistics 12, 5789–5840, 2015 5 and Applied Probability, Chapman and Hall, London, 26, mainly chapters 3 and 4, 1986. 5806 Simmons, A., Uppala, S., Dee, D., and Kobayashi, S.: ERA-Interim: new ECMWF reanalysis South Asia river flow products from 1989 onwards, ECMWF, Shinfield, Reading, UK, Newsletter – Winter 2006/07, projections and their 110, 0–11, 2007. 5793, 5795, 5797 implications for water 10 Singh, O. and Kumar, M.: Flood events, fatalities and damages in India from 1978 to 2006, Nat. resources Hazards, 69, 1815–1834, 2013. 5814 Smakhtin, V. and Anputhas, M.: An assessment of environmental flow requirements of Indian C. Mathison et al. river basins, IWMI Research Report 107, available at: http://www.iwmi.cgiar.org/Publications/ IWMI_Research_Reports/PDF/PUB107/RR107.pdf, last access: October 2014, 2006. 5812 15 Smakhtin, V. U., Shilpakar, R. L., and Hughes, D. A.: Hydrology-based assessment Title Page of environmental flows: an example from Nepal, Hydrolog. Sci. J., 51, 207–222, doi:10.1623/hysj.51.2.207, 2006. 5812 Abstract Introduction Thenkabail, P. S., Schull, M., and Turral, H.: Ganges and Indus river basin land use/land cover LULC) and irrigated area mapping using continuous streams of MODIS data, Remote Sens. Conclusions References 20 Environ., 95, 317–341, doi:10.1016/j.rse.2004.12.018, 2005. 5795 Tables Figures Tuinenburg, O. A., Hutjes, R. W. A., Stacke, T., Wiltshire, A., and Lucas-Picher, P.: Effects of irrigation in india on the atmospheric water budget, J. Hydrometeorol., 15, 1028–1050, doi:10.1175/JHM-D-13-078.1, 2014. 5799, 5813 J I United Nations: Fertility Levels and Trends as Assessed in the 2012 Revision of World J I 25 Population Prospects, Department of Economic and Social Affairs, Population Division, available at: http://esa.un.org/unpd/wpp/index.htm, last access: August 2014, 2013. 5791, Back Close 5815 Vörösmarty, C., Fekete, B., Meybeck, M., and Lammers, R.: Geomorphometric attributes Full Screen / Esc of the global system of rivers at 30 min spatial resolution, J. Hydrol., 237, 17–39, 30 doi:10.1016/S0022-1694(00)00282-1, 2000a. 5798 Printer-friendly Version Vörösmarty, C., Fekete, B., Meybeck, M., and Lammers, R.: Global system of rivers: its role in organizing continental land mass and defining land-to-ocean linkages, Global Biogeochem. Interactive Discussion Cy., 14, 599–621, doi:10.1029/1999GB900092, 2000b. 5798 5825 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion Vörösmarty, C. J., McIntyre, P. B., Gessner, M. O., Dudgeon, D., Prusevich, A., Green, P., Glidden, S., Bunn, S. E., Sullivan, C. A., Liermann, C. R., and Davies, P. M.: Global threats to HESSD human water security and river biodiversity, Nature, 467, 555–561, doi:10.1038/nature09440, 2010. 5811 12, 5789–5840, 2015 5 Wescoat Jr., J.: Managing the Indus River basin in light of climate change: four conceptual approaches, Global Environ. Chang., 1, 381–395, doi:10.1016/0959-3780(91)90004-D, 1991. 5795 South Asia river flow Wiltshire, A. J.: Climate change implications for the glaciers of the Hindu Kush, Karakoram and projections and their Himalayan region, The Cryosphere, 8, 941–958, doi:10.5194/tc-8-941-2014, 2014. 5793, implications for water 10 5794 resources Yatagai, A., Kamiguchi, K., Arakawa, O., Hamada, A., Yasutomi, N., and Kitoh, A.: Aphrodite: constructing a long-term daily gridded precipitation dataset for asia based on a dense C. Mathison et al. network of rain gauges., B. Am. Meteorol. Soc., 93, 1401–1415, doi:10.1175/BAMS-D-11- 00122.1, 2012. 5794, 5795 15 Yoshida, S.: Fundamentals of rice crop science, Manila, Philipines, available at: http://books. Title Page irri.org/9711040522_content.pdf, last access: October 2014, 1981. 5815 Abstract Introduction

Conclusions References

Tables Figures

J I

Back Close

Full Screen / Esc

Printer-friendly Version

Interactive Discussion

South Asia river ﬂow Table 1. Table listing the rivers and gauges (including their location) used in this analysis; all projections and their the observations shown here are from GRDC. The abbreviations used in Fig.1 are given in implications for water column one. The Years of data column includes the number of years that data is available since resources 1950 with c to denote where data is continuous and u to show where the data is available for that number of years but not as a continuous dataset. C. Mathison et al.

Map abbreviation River name Gauge name Latitude Longitude Years of data IND_KOT Indus Kotri 25.37 68.37 14u (1950–1978) Title Page IND_ATT Indus Attock 33.9 72.25 6c (1973–1979) Abstract Introduction CHE_PAN Chenab Panjnad 29.35 71.03 6c (1973–1979) BHA_TEH Bhagirathi Tehri Dam 30.4 78.5 3c (2001–2004) Conclusions References KAR_BEN Karnali River Benighat 28.96 81.12 25u (1963–1993) KAR_CHI Karnali River Chisapani 28.64 81.29 31c (1962–1993) Tables Figures NAR_DEV Narayani Devghat 27.71 84.43 23u (1963–1993) ARU_TUR Arun Turkeghat 27.33 87.19 10c (1976–1986) J I GAN_FAR Ganges Farakka 25.0 87.92 18u (1950–1973) BRA_BAH Brahmaputra Bahadurabad 25.18 89.67 12u (1969–1992) J I BRA_YAN Brahmaputra Yangcun 29.28 91.88 21u (1956–1982) Back Close BRA_PAN Brahmaputra Pandu 26.13 91.7 13u (1956–1979) Full Screen / Esc

Printer-friendly Version

Interactive Discussion

South Asia river ﬂow Table 2. Table showing the average percentage change for the two models in the number of projections and their times the modelled river ﬂow is less than the 10th percentile and greater than the 90th percentile implications for water of the 1970–2000 period for the 2050s and 2080s future periods. resources River Gauge < 10th percentile % change > 90th percentile % change C. Mathison et al. 2050s 2080s 2050s 2080s Indus Kotri −55.4 −89.2 60.8 55.4 Indus Attock −70.3 −95.9 70.3 81.1 Title Page Karnali River Benighat −39.2 −73.0 63.5 81.1 Abstract Introduction Karnali River Chisapani −27.0 −56.8 60.8 79.7 Narayani Devghat −21.6 −54.1 75.7 110.8 Conclusions References Arun Turkeghat −63.5 −90.5 66.2 116.2 Brahmaputra Yangcun 0 0 20.3 36.5 Tables Figures Brahmaputra Pandu −59.5 −79.7 47.3 113.5 Brahmaputra Bahadurabad −48.6 −64.9 67.6 114.9 J I Ganges Farakka −36.5 −52.7 68.9 102.7 ∗ Bhagirathi Tehri Dam −4.1 12.2 13.5 41.9 J I Chenab Panjnad −58.1 −83.8 43.2 50.0 Back Close

∗ This value is the only positive value in the table. Full Screen / Esc

Printer-friendly Version

Interactive Discussion

South Asia river flow projections and their implications for water Table 3. Table of implications of changes in water resources. resources Types of change Implications for water resource Adaptation options Other issues Large annual Abundance some years and Building storage capacity Type of water storage is C. Mathison et al. variability scarcity in others make e.g. rainwater harvesting. important e.g. reservoirs/dams it difficult to plan budgets Improvement of irrigations systems. have both political for different users. Development of water efficient, and ecological implications. high yielding crop varieties. Developing new crops takes time. Title Page Changes in Increases in peak flows could be Improving river channel capacity. Flood protection levels peak flow – timing positive for irrigation and Diverting excess water to a different valley. do not match demographic and magnitude domestic supply but could increase Storing the excess water for low flow periods trends so vulnerability Abstract Introduction the risk of flooding. e.g. through rainwater harvesting. to flooding remains high Peak flows occurring later and/or Improving drainage and water recycling. in this region Conclusions References decreases in peak flows could reduce Adopting varieties of crops that grow (Gupta et al., 2003). availability of water for irrigation when water for irrigation is more Market development for at crucial crop development stages readily available new crops takes time Tables Figures negatively impacting yields. Changes in low Increases in the magnitude of the low Adaptations to avoid flooding during flows – timing flows could be positive for irrigation peak flow periods could provide resource J I and magnitude and domestic supply. during low flow periods. Decreases could mean less resource Development of water efficient, available for irrigation high yielding J I leading to reduced yields crop varieties Back Close

Full Screen / Esc

Printer-friendly Version

Interactive Discussion

South Asia river ﬂow projections and their implications for water resources

C. Mathison et al.

Title Page

Abstract Introduction

Conclusions References

Tables Figures

J I

Back Close

Figure 1. A map showing the locations of the river gauges used in this analysis. Full Screen / Esc

Printer-friendly Version

Interactive Discussion

South Asia river ﬂow projections and their implications for water resources

C. Mathison et al.

Title Page

Abstract Introduction

Conclusions References

Tables Figures

J I

Back Close

Full Screen / Esc

Printer-friendly Version Figure 2. The spatial distribution of the seasonal mean total precipitation for the monsoon Interactive Discussion period (June, July, August, September) for APHRODITE observations (top left), ERAint (top right), HadCM3 (bottom left) and ECHAM5 (bottom right). 5831 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion HESSD 12, 5789–5840, 2015

South Asia river ﬂow

Annual average RCM Total precipitation projections and their Indus Ganges/Brahmaputra 3.0 6.5 implications for water

6.0 resources

2.5 5.5 C. Mathison et al.

5.0 2.0

4.5

1.5 Title Page 4.0 Total precipitation mm/day Total precipitation mm/day 3.5 Abstract Introduction 1.0

3.0 Conclusions References

0.5 2.5 1963 1983 2003 2023 2043 2063 2083 1963 1983 2003 2023 2043 2063 2083 Tables Figures year year

Aphrodite_obs ERAint HadCM3 ECHAM5 Aphrodite_obs smooth ERAint smooth HadCM3 smooth ECHAM5 smooth J I

Figure 3. Annual mean total precipitation for the Indus (left) and Ganges/Brahmaputra (right) J I catchments for each model run (HadCM3 – red, ECHAM5 – blue, ERAint – cyan lines) plotted against APHRODITE observations (black line). Back Close Full Screen / Esc

Printer-friendly Version

Interactive Discussion

South Asia river ﬂow projections and their 1971-2000 RCM Total precipitation Indus Ganges/Brahmaputra 4.0 12 implications for water

3.5 resources 10 3.0 C. Mathison et al. 8 2.5

2.0 6 Title Page 1.5 4

Total precipitation mm/day Total precipitation mm/day Abstract Introduction 1.0

2 0.5 Conclusions References

0.0 0 Jan Mar May Jul Sep Nov Jan Mar May Jul Sep Nov Tables Figures Time / Month Time / Month

Aphrodite_obs ERAint HadCM3 ECHAM5 J I Figure 4. Seasonal cycle of total precipitation for the Indus (left) and Ganges/Brahmaputra J I (right) catchments for each model run (HadCM3 – red, ECHAM5 – blue, ERAint – cyan lines) plotted against APHRODITE observations (black line). Back Close

Full Screen / Esc

Printer-friendly Version

Interactive Discussion

3 Annual RCM INDUS_KOTRI 3 Annual RCM INDUS_ATTOCK 18 8 m m 16 7 0 0

0 14 0 6 0 0 HESSD 12 1 1 5 10 e e

t t 4 8 a a

r r 3 6

w 4 w 2 o o

l l 12, 5789–5840, 2015

f 2 f 1

r 0 r 0 e 1960 1980 2000 2020 2040 2060 2080 2100 e 1960 1980 2000 2020 2040 2060 2080 2100 v v i i

R Time / Year R Time / Year 1 1 − − s s

3 Annual RCM CHENAB_PANJNAD 3 Annual RCM BHAGIRATHI_TEHRI 7 0.7 m m

0 6 0 0.6

0 0 South Asia river ﬂow

0 5 0 0.5 1 1 4 0.4 e e t t

a 3 a 0.3 r r

projections and their 2 0.2 w w o o

l 1 l 0.1 f f

r 0 r 0.0 e 1960 1980 2000 2020 2040 2060 2080 2100 e 1960 1980 2000 2020 2040 2060 2080 2100 implications for water v v i i

R Time / Year R Time / Year 1 1 − −

s s resources

3 Annual RCM KARNALI_BENIGHAT 3 Annual RCM KARNALI_CHISAPANI 2.0 3.5 m m

0 0 3.0

0 1.5 0

0 0 2.5 1 1 2.0 e e t 1.0 t C. Mathison et al.

a a 1.5 r r 1.0

w 0.5 w o o

l l 0.5 f f

r 0.0 r 0.0 e 1960 1980 2000 2020 2040 2060 2080 2100 e 1960 1980 2000 2020 2040 2060 2080 2100 v v i i

R Time / Year R Time / Year 1 1 − − s s

3 Annual RCM NARAYANI_DEVGHAT 3 Annual RCM ARUN_TURKEGHAT 2.5 0.8 m m 0.7 Title Page 0 2.0 0 0 0 0.6 0 0

1 1 0.5 1.5 e e

t t 0.4

a 1.0 a r r 0.3 Abstract Introduction

w w 0.2 0.5 o o l l

f f 0.1

r 0.0 r 0.0 e 1960 1980 2000 2020 2040 2060 2080 2100 e 1960 1980 2000 2020 2040 2060 2080 2100 v v i i R Time / Year R Time / Year Conclusions References 1 1 − − s s

3 Annual RCM GANGES_FARAKKA 3 Annual RCM BRAHMAPUTRA_BAHADURABAD 30 60 m m

0 25 0 50 0 0

0 20 0 40 Tables Figures 1 1

e e

t 15 t 30 a a r r

10 20 w w

o 5 o 10 l l f f

r 0 r 0 e 1960 1980 2000 2020 2040 2060 2080 2100 e 1960 1980 2000 2020 2040 2060 2080 2100 v v i i

R Time / Year R Time / Year J I 1 1 − − s s

3 Annual RCM BRAHMAPUTRA_YANGCUN 3 Annual RCM BRAHMAPUTRA_PANDU 3.5 35 m m

0 3.0 0 30

0 0 J I

0 2.5 0 25 1 1 2.0 20 e e t t

a 1.5 a 15 r r 1.0 10

w w Back Close o o

l 0.5 l 5 f f

r 0.0 r 0 e 1960 1980 2000 2020 2040 2060 2080 2100 e 1960 1980 2000 2020 2040 2060 2080 2100 v v i i R Time / Year R Time / Year Full Screen / Esc

Observed HadCM3 ECHAM5 ERAint HadCM3 smooth ECHAM5 smooth Printer-friendly Version Figure 5. Timeseries of river ﬂows showing available observations (black) and RCM runs Interactive Discussion (HadCM3 – red, ECHAM5 – blue, ERAint – cyan lines) from 1971–2100. Paler lines are annual averages and darker lines are a rolling smoothed average. 5834 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion 1 1 − − s s

3 1971-2000 RCM INDUS_KOTRI 3 1971-2000 RCM INDUS_ATTOCK 30 12 m m HESSD

0 25 0 10 0 0

0 20 0 8 1 1 15 6 e e t t

a 10 a 4 r r 5 2 12, 5789–5840, 2015 w w o o

l 0 l 0 f f

r 5 r 2 e Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec e Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec v v i i

R Time / Month R Time / Month 1 1 − − s s

3 1971-2000 RCM CHENAB_PANJNAD 3 1971-2000 RCM BHAGIRATHI_TEHRI 14 2.0 m m 12 South Asia river ﬂow 0 0 1.5 0 10 0 0 0

1 8 1 1.0 e e t 6 t projections and their a a 0.5 r 4 r

w 2 w 0.0 o o l l

f 0 f

r r implications for water 2 0.5 e Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec e Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec v v i i

R Time / Month R Time / Month

1 1 resources − − s s

3 1971-2000 RCM KARNALI_BENIGHAT 3 1971-2000 RCM KARNALI_CHISAPANI 3.5 7 m m 3.0 6 0 0

0 2.5 0 5 0 0

1 2.0 1 4

e e

t 1.5 t 3 C. Mathison et al. a a

r 1.0 r 2

w 0.5 w 1 o o l l

f 0.0 f 0

r 0.5 r 1 e Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec e Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec v v i i

R Time / Month R Time / Month 1 1 − − s s

3 1971-2000 RCM NARAYANI_DEVGHAT 3 1971-2000 RCM ARUN_TURKEGHAT 6 2.0 m m Title Page

0 5 0 1.5 0 0

0 4 0 1 1 3 1.0 e e t t

a 2 a 0.5 r r 1

w w Abstract Introduction 0.0 o o

l 0 l f f

r 1 r 0.5 e Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec e Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec v v i i

R Time / Month R Time / Month

1 1 Conclusions References − − s s

3 1971-2000 RCM GANGES_FARAKKA 3 1971-2000 RCM BRAHMAPUTRA_BAHADURABAD 60 100 m m

0 50 0 80 0 0

0 40 0 1 1

60 Tables Figures e e

t 30 t

a a 40 r r

w w 20 o 10 o l l f f

r 0 r 0 e Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec e Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec v v i i R Time / Month R Time / Month J I 1 1 − − s s

3 1971-2000 RCM BRAHMAPUTRA_YANGCUN 3 1971-2000 RCM BRAHMAPUTRA_PANDU 12 50 m m

0 10 0 40 0 0

0 8 0

1 1 J I 6 30 e e t t

a 4 a 20 r r 2 w w 10 o o

l 0 l f f

Back Close r 2 r 0 e Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec e Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec v v i i

R Time / Month R Time / Month Full Screen / Esc Observed HadCM3 1971-2000 1.5 stdev HadCM3 ERAint ECHAM5 1971-2000 1.5 stdev ECHAM5

Printer-friendly Version Figure 6. Seasonal cycle of river ﬂow at individual river gauges; observed (black solid line) and for each of the RCMs (HadCM3 – red, ECHAM5 – blue, ERAint – cyan lines) for 1971–2000; Interactive Discussion with shaded regions showing 1.5 SD from the mean for the two simulations for the same period.

3 Future RCM INDUS_KOTRI 3 Future RCM INDUS_ATTOCK 30 12 m m

0 25 0 10 0 0

0 20 0 8 HESSD 1 1 15 6 e e t t

a 10 a 4 r r 5 2 w w o o

l 0 l 0 12, 5789–5840, 2015 f f

r 5 r 2 e Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec e Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec v v i i

R Time / Month R Time / Month 1 1 − − s s

3 Future RCM CHENAB_PANJNAD 3 Future RCM BHAGIRATHI_TEHRI 14 2.5 m m 12 0 0 2.0 0 10 0 South Asia river ﬂow 0 0 1.5 1 8 1

e e

t 6 t 1.0 a a

r 4 r 0.5 projections and their

w 2 w

o o 0.0 l l

f 0 f

r 2 r 0.5 e Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec e Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec implications for water v v i i

R Time / Month R Time / Month 1 1 − −

s s resources

3 Future RCM KARNALI_BENIGHAT 3 Future RCM KARNALI_CHISAPANI 3.5 8 m m 3.0 7 0 0

0 2.5 0 6 0 0 5 1 2.0 1 4 e e

t 1.5 t 3 C. Mathison et al. a a

r 1.0 r 2

w 0.5 w 1 o o l l

f 0.0 f 0

r 0.5 r 1 e Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec e Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec v v i i

R Time / Month R Time / Month 1 1 − − s s

3 Future RCM NARAYANI_DEVGHAT 3 Future RCM ARUN_TURKEGHAT 6 2.0 m m

Title Page

0 5 0 1.5 0 0

0 4 0 1 1 3 1.0 e e t t

a 2 a 0.5 r r

Abstract Introduction 1 w w 0.0 o o

l 0 l f f

r 1 r 0.5 e Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec e Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec v v i i

R Time / Month R Time / Month Conclusions References 1 1 − − s s

3 Future RCM GANGES_FARAKKA 3 Future RCM BRAHMAPUTRA_BAHADURABAD 60 100 m m

0 50 0 80 0 0 Tables Figures 0 40 0 1 1

60 e e

t 30 t

a a 40 r r

w w 20 o 10 o l l f f

r 0 r 0 e Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec e Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec v v

i i J I

R Time / Month R Time / Month 1 1 − − s s

3 Future RCM BRAHMAPUTRA_YANGCUN 3 Future RCM BRAHMAPUTRA_PANDU 12 50 m m

0 10 0 J I 40 0 0

0 8 0 1 1 6 30 e e t t

a 4 a 20 r r 2 Back Close w w 10 o o

l 0 l f f

r 2 r 0 e Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec e Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec v v i i

R Time / Month R Time / Month Full Screen / Esc

HadCM32040-2070 ECHAM52040-2070 1971-2000 1.5 stdev HadCM3 HadCM32068-2098 ECHAM52068-2098 1971-2000 1.5 stdev ECHAM5 Printer-friendly Version Figure 7. Seasonal cycle of river ﬂow in each of the RCMs (HadCM3 – red, ECHAM5 – blue) Interactive Discussion for the two future periods: 2050s (solid lines) and 2080s (dashed lines), with shaded regions showing 1.5 SD from the mean for 1971–2000 for each river gauge. 5836 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion

RCM INDUS_KOTRI RCM INDUS_ATTOCK 0.00010 0.00025 0.00008 0.00020 HESSD 0.00006 0.00015 0.00004 0.00010 Density Density 0.00002 0.00005 12, 5789–5840, 2015 0.00000 0.00000 0 12 24 36 48 60 0 4 8 12 16 20 3 1 3 1 River flow rate 1000 m s− River flow rate 1000 m s− RCM CHENAB_PANJNAD RCM BHAGIRATHI_TEHRI 0.00025 0.0030 0.00020 0.0025 0.0020 South Asia river ﬂow 0.00015 0.0015 0.00010 0.0010 Density Density projections and their 0.00005 0.0005 0.00000 0.0000 0.0 4.6 9.2 13.8 18.4 23.0 0 1 2 3 4 5 implications for water 3 1 3 1 River flow rate 1000 m s− River flow rate 1000 m s− RCM KARNALI_BENIGHAT RCM KARNALI_CHISAPANI 0.0007 0.00040 resources 0.0006 0.00035 0.0005 0.00030 0.00025 0.0004 0.00020 0.0003 0.00015 C. Mathison et al. Density 0.0002 Density 0.00010 0.0001 0.00005 0.0000 0.00000 0.0 1.6 3.2 4.8 6.4 8.0 0.0 3.2 6.4 9.6 12.8 16.0 3 1 3 1 River flow rate 1000 m s− River flow rate 1000 m s− RCM NARAYANI_DEVGHAT RCM ARUN_TURKEGHAT 0.0005 0.0014 0.0012 Title Page 0.0004 0.0010 0.0003 0.0008 0.0002 0.0006

Density Density 0.0004 Abstract Introduction 0.0001 0.0002 0.0000 0.0000 0.0 2.2 4.4 6.6 8.8 11.0 0.00 0.62 1.24 1.86 2.48 3.10 3 1 3 1 River flow rate 1000 m s− River flow rate 1000 m s− Conclusions References RCM GANGES_FARAKKA RCM BRAHMAPUTRA_BAHADURABAD 0.00005 0.000025 0.00004 0.000020 Tables Figures 0.00003 0.000015 0.00002 0.000010 Density Density 0.00001 0.000005 0.00000 0.000000 0 18 36 54 72 90 0 40 80 120 160 200 3 1 3 1 J I River flow rate 1000 m s− River flow rate 1000 m s− RCM BRAHMAPUTRA_YANGCUN RCM BRAHMAPUTRA_PANDU 0.00030 0.000035 0.00025 0.000030 J I 0.00020 0.000025 0.000020 0.00015 0.000015 0.00010 Density Density 0.000010 Back Close 0.00005 0.000005 0.00000 0.000000 0.0 2.6 5.2 7.8 10.4 13.0 0 18 36 54 72 90 3 1 3 1 River flow rate 1000 m s− River flow rate 1000 m s− Full Screen / Esc

1971-2000:HadCM3 2040-2070:HadCM3 2068-2098:HadCM3 1971-2000:ECHAM5 2040-2070:ECHAM5 2068-2098:ECHAM5 Printer-friendly Version Figure 8. The distribution of the river ﬂow in the HadCM3 and ECHAM5 (HadCM3 – red, Interactive Discussion ECHAM5 – blue) runs for three periods: historical (1971–2000 – solid lines) and two future periods (2050s – dashed lines and 2080s – dotted lines) plotted as a pdf for each river gauge. 5837 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion

GANGES_FARAKKA Years where riverflow<10th %ile for 1971-2000 HESSD 1 −

s HadCM3 HadCM3 37 3 ECHAM5 ECHAM5 37 m 12, 5789–5840, 2015

2.0 0 0 0 1

t 1.5 a r

South Asia river ﬂow w o l f

1.0 projections and their r e v i implications for water R 1965 1970 1975 1980 1985 1990 1995 2000 2005 Years with monthly riverflows<10th percentile resources

Years where riverflow<10th %ile for 1971-2000 2.4 C. Mathison et al. 1 −

s HadCM3 HadCM3 16

3 2.2 ECHAM5 ECHAM5 31 m 2.0 0 0

0 1.8 1

e Title Page

t 1.6 a r 1.4 w o l 1.2 Abstract Introduction f

e 1.0 v i

R 0.8 Conclusions References 2035 2040 2045 2050 2055 2060 2065 2070 2075 Years with monthly riverflows<10th percentile Tables Figures Years where riverflow<10th %ile for 1971-2000 2.4 1 −

s HadCM3 HadCM3 9 3 2.2 ECHAM5 ECHAM5 26 m

J I 0

0 2.0 0 1

e J I

t 1.8 a r

w 1.6 Back Close o l f

r 1.4 e v i Full Screen / Esc R 1.2 2065 2070 2075 2080 2085 2090 2095 2100 Years with monthly riverflows<10th percentile Printer-friendly Version Figure 9. Comparison of the lowest 10 % of river ﬂows at the Farakka barrage on the Ganges Interactive Discussion river against the 10th percentile for the 1971–2000 period for 1971–2000 (top), 2050s (middle) and 2080s (bottom) for HadCM3 (red triangles) and ECHAM5 (blue stars). 5838 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion

GANGES_FARAKKA Years where riverflow>90th %ile for 1971-2000 HESSD 70 1 −

s HadCM3 HadCM3 37

3 65 ECHAM5 ECHAM5 37 m 12, 5789–5840, 2015

0 60 0 0 1

55 e t

a 50 r

South Asia river ﬂow w

o 45 l f

projections and their r

e 40 v i implications for water R 35 1970 1975 1980 1985 1990 1995 2000 2005 Years with monthly riverflows>90th percentile resources

Years where riverflow>90th %ile for 1971-2000 75 C. Mathison et al. 1 −

s HadCM3 HadCM3 60

3 70 ECHAM5 ECHAM5 65 m 65 0 0

0 60 1

e Title Page

t 55 a r 50 w o l 45 Abstract Introduction f

e 40 v i

R 35 Conclusions References 2035 2040 2045 2050 2055 2060 2065 2070 2075 Years with monthly riverflows>90th percentile Tables Figures Years where riverflow>90th %ile for 1971-2000 100 1 −

s HadCM3 HadCM3 71

3 90 ECHAM5 ECHAM5 79 m

J I

0 80 0 0 1

e J I t

a 60 r

w Back Close o 50 l f

e 40 v i Full Screen / Esc R 30 2065 2070 2075 2080 2085 2090 2095 2100 Years with monthly riverflows>90th percentile Printer-friendly Version Figure 10. Comparison of the highest 10 % of river ﬂows at the Farakka barrage on the Ganges Interactive Discussion river against the 90th percentile for the 1971–2000 period for 1971–2000 (top), 2050s (middle) and 2080s (bottom) for HadCM3 (red triangles) and ECHAM5 (bluen stars). 5839 icsinPpr|Dsuso ae icsinPpr|Dsuso ae | Paper Discussion | Paper Discussion | Paper Discussion | Paper Discussion

90th Percentile per decade 1 1

− INDUS_KOTRI − INDUS_ATTOCK s s 3 28 3 12 HESSD m m 26 11 0 0

0 24 0 0 0 10 1 1 22 e e

t t 9

a 20 a 12, 5789–5840, 2015 r r 8

w 18 w o o l l

f 16 f 7

r r

e 14 e 6 v v

i 1970s 1980s 1990s 2000s 2010s 2020s 2030s 2040s 2050s 2060s 2070s 2080s 2090s i 1970s 1980s 1990s 2000s 2010s 2020s 2030s 2040s 2050s 2060s 2070s 2080s 2090s R Time / Decade R Time / Decade 1 1

− CHENAB_PANJNAD − BHAGIRATHI_TEHRI s s South Asia river ﬂow

3 14 3 2.0 m m

13 1.8 0 0

0 12 0 1.6

0 0 projections and their

1 11 1 1.4

e e

t 10 t 1.2 a a

r 9 r 1.0

w w implications for water 8 0.8 o o l l f f

7 0.6 r r

e 6 e 0.4 v v i 1970s 1980s 1990s 2000s 2010s 2020s 2030s 2040s 2050s 2060s 2070s 2080s 2090s i 1970s 1980s 1990s 2000s 2010s 2020s 2030s 2040s 2050s 2060s 2070s 2080s 2090s resources R Time / Decade R Time / Decade 1 1

− KARNALI_BENIGHAT − KARNALI_CHISAPANI s s

3 3.2 3 7.0 m m 3.0 6.5 C. Mathison et al. 0 0

0 2.8 0 6.0 0 0 1 1 2.6 5.5 e e t t

a 2.4 a 5.0 r r

w 2.2 w 4.5 o o l l

f 2.0 f 4.0

r r

e 1.8 e 3.5 v v

i 1970s 1980s 1990s 2000s 2010s 2020s 2030s 2040s 2050s 2060s 2070s 2080s 2090s i 1970s 1980s 1990s 2000s 2010s 2020s 2030s 2040s 2050s 2060s 2070s 2080s 2090s R Time / Decade R Time / Decade Title Page 1 1

− NARAYANI_DEVGHAT − ARUN_TURKEGHAT s s

3 6.0 3 2.0 m m 5.5 1.9 0 0 Abstract Introduction 0 0 1.8

0 5.0 0

1 1 1.7

e e

t 4.5 t 1.6 a a

r r 1.5 4.0

w w 1.4 o o

l l Conclusions References

f 3.5 f

1.3 r r

e 3.0 e 1.2 v v

− GANGES_FARAKKA − BRAHMAPUTRA_BAHADURABAD s s

3 65 3 90 m m

60 85 0 0

0 55 0 80 0 0

1 1 75 50 e e

t t 70

a 45 a r r 65 J I

w 40 w 60 o o l l

f 35 f 55 r r

e 30 e 50 v v

i 1970s 1980s 1990s 2000s 2010s 2020s 2030s 2040s 2050s 2060s 2070s 2080s 2090s i 1970s 1980s 1990s 2000s 2010s 2020s 2030s 2040s 2050s 2060s 2070s 2080s 2090s R Time / Decade R Time / Decade J I 1 1

− BRAHMAPUTRA_YANGCUN − BRAHMAPUTRA_PANDU s s

3 8.5 3 50 m m 8.0 Back Close 0 0

0 0 45

0 7.5 0 1 1

e e

t 7.0 t 40 a a r r 6.5

w w 35 Full Screen / Esc o o l l

f 6.0 f

r r

e 5.5 e 30 v v

i 1970s 1980s 1990s 2000s 2010s 2020s 2030s 2040s 2050s 2060s 2070s 2080s 2090s i 1970s 1980s 1990s 2000s 2010s 2020s 2030s 2040s 2050s 2060s 2070s 2080s 2090s R Time / Decade R Time / Decade

HadCM3 ECHAM5 Printer-friendly Version

Interactive Discussion Figure 11. The 90th percentile of river ﬂow for each decade for HadCM3 (red triangles) and ECHAM5 (blue circles) for each river gauge. 5840